Differential Effects on Student Demographic Groups of Using ACT College Readiness Assessment Composite Score, ACT Benchmarks, and High School Grade Point Average for Predicting Long-Term College Success through Degree Completetion

*050205130*

Rev 1

Differential Effects on Student Demographic

Groups of Using ACT

College Readiness

Assessment Composite Score, ACT

Benchmarks, and High School Grade Point

Average for Predicting Long-Term College

Success through Degree Completion

Justine Radunzel

Julie Noble

August 2013

ACT Research

Report Series

2013 (5)

For additional copies:

ACT Research Report Series

P.O. Box 168

Iowa City, IA 52243-0168

Differential Effects on Student Demographic Groups of

Using ACT

College Readiness Assessment Composite

Score, ACT Benchmarks, and High School Grade Point

Average for Predicting Long-Term College Success through

Degree Completion

Justine Radunzel

Julie Noble

Abstract

In this study, we evaluated the differential effects on racial/ethnic, family income, and

gender groups of using ACT

College Readiness Assessment Composite score and high school

grade point average (HSGPA) for predicting long-term college success. Outcomes included

annual progress towards a degree (based on cumulative credit-bearing hours earned), degree

completion, and cumulative grade point average at 150% of normal time to degree completion

(year 6 and year 3 for four- and two-year institutions, respectively). We also evaluated the utility

of the individual ACT College Readiness Benchmarks for predicting college success for each

demographic group.

Data for this study included over 190,000 ACT-tested students who enrolled in college as

first-time entering students in fall, 2000 through 2006. Over 100 total two- and four-year

institutions were represented. We used hierarchical logistic models to estimate institution-

specific probabilities of college success for all students and each demographic group based on

their ACT test scores and HSGPA. Accuracy and success rates for each student group were

calculated at total-group optimal selection values using the distributions of ACT Composite

score and HSGPA for each institution’s approximate applicant pool; these rates were then

summarized across institutions. Results were disaggregated by institution type.

Total-group predictions based on ACT Composite score generally overestimated the

long-term college success of underrepresented minority students (by, at most, 0.11 across

outcomes), lower-income students (by, at most, 0.07), and male students (by, at most, 0.13) and,

to a lesser extent, underestimated the success of White students (by, at most, 0.04), higher-

income students (by, at most, 0.07), and female students (by, at most, 0.10). The degree of

differential prediction by gender was less pronounced for the progress to degree and degree

iii

completion outcomes than for achieving levels of year 6/year 3 cumulative grade point average

(GPA). There was minimal differential prediction by family income for achieving levels of year

6/year 3 cumulative GPA. For racial/ethnic and family income groups, there was greater over-

and underprediction associated with using HSGPA than with using ACT Composite score. The

opposite was true for gender. Differential prediction by student demographic groups was also

observed at the ACT College Readiness Benchmark scores with the direction of the differential

prediction being consistent with that observed when ACT Composite score and/or HSGPA was

used.

For each student demographic group, test scores increased prediction accuracy over that

for HSGPA. Typical percentages of correct classifications at total-group optimal selection values

were generally higher for underrepresented minority and lower-income students than for White

and higher-income students; these percentages were similar for female and male students.

Contrary to prior claims made, results from this study suggest that minority and lower-

income students are not disadvantaged by using ACT Composite score or the ACT Benchmark

scores to predict long-term college success. This finding held across multiple college outcomes

at both two- and four-year institutions.

Acknowledgments

The authors thank Edgar Sanchez and Richard Sawyer for their helpful comments and

suggestions on an earlier draft of this report.

Differential Effects on Student Demographic Groups of Using ACT

College Readiness

Assessment Composite Score, ACT Benchmarks, and High School Grade Point Average for

Predicting Long-Term College Success through Degree Completion

Introduction

To meet their admission goals while at the same time fulfilling their educational mission

of maintaining equal opportunity and diversity in student enrollments, four-year postsecondary

institutions often use multiple measures, including both academic and non-academic ones, in

determining the likelihood that a student will be successful in college (Clinedinst, Hurley, &

Hawkins, 2011). Academic measures often include grades in college preparatory courses,

strength of high school curriculum, standardized test scores (ACT or SAT), and high school

grade point average (HSGPA), because these measures have been found to identify accurately

students who are ready for college and to predict students’ eventual success in college. One

outcome that is commonly used by institutions for helping them make admission decisions is

first-year academic performance, as measured by first-year college grade point average (GPA).

But, due to the increased pressure that institutions are currently under to improve graduation

rates, institutions are also considering outcomes beyond the first year of college, including

evaluating the likelihood that applicants will complete a degree within six years (Higher

Education Research Institute, 2011; Saupe & Curs, 2008).

Two-year institutions are also feeling the pressure to increase graduation rates. And, even

though most two-year institutions currently practice open admissions, about one-fifth of them

use standardized test scores or HSGPA as part of their admission process (Breland, Maxey,

Gernand, Cumming, & Trapani, 2002). Moreover, due to the reduced resources available to them

some two-year institutions are having to prioritize access; restrict enrollment; eliminate lower-

level, developmental courses; and identify students who are likely to graduate or transfer to a

four-year institution (González, 2012). In addition, two-year institutions are being encouraged to

evaluate intermediate outcomes that measure progress towards degree completion to help

determine the reasons why so many students are not completing degrees (Moore, Shulock, &

Offenstein, 2009). Two-year institutions also use students’ test scores and HSGPAs to counsel

their applicants, including those who appear to be at risk of not succeeding in college (Habley,

Valiga, McClanahan, & Burkum, 2010).

In light of the push for increased accountability in higher education and the growing

concerns for open access remaining the norm at two-year institutions (González, 2012), we

recently evaluated the use of ACT Composite (ACTC) score and HSGPA for identifying

students who are likely to be successful in college beyond the first year for both four- and two-

year institutions (Radunzel & Noble, 2012a). In this study, we found that the typical percentages

of correct classifications at optimal ACTC scores for progressing towards and completing a

degree were moderately high (64% to 71% at four-year institutions and 65% to 77% at two-year

institutions). Across the college outcomes considered in the study, using ACTC score and

HSGPA in combination resulted in greater prediction accuracy, and was more effective for

identifying successful students among those expected to be successful, relative to using either

pre-enrollment achievement measure separately. Other researchers (Schmitt, Keeney, Oswald,

Pleskac, Billington, Sinha, & Zorzie, 2009) have also reported relatively high percentages of

correct classifications when predicting bachelor’s degree completion using SAT/ACT scores and

HSGPA jointly (for 63% of the students). In addition, it has been shown that college success

rates (including six-year bachelor’s degree completion rates) are substantially greater for

students with higher ACTC scores and HSGPAs than for those with lower scores or HSGPAs

(Radunzel & Noble, 2012b).

In our earlier study (Radunzel & Noble, 2012a), we also investigated the usefulness of

the ACT College Readiness Benchmarks in each of the subject areas for predicting long-term

college success. The ACT Benchmarks are the minimum ACT test scores required for students to

have a high probability of success in first-year, credit-bearing college courses–English

Composition, College Algebra, social sciences courses, and Biology (Allen & Sconing, 2005),

and provide an empirical definition of college readiness. The ACT Benchmarks were identified

as the typical scores across both two- and four-year institutions that maximized the accuracy for

predicting success (defined as earning a grade of B or higher) in the corresponding courses.

Meeting the ACT Benchmarks has also been shown to be positively associated with early and

long-term college success, such as enrolling and persisting in college and completing a degree

(Radunzel & Noble, 2012b; ACT, 2010a). Results from Radunzel and Noble (2012a) are

consistent with these other findings, and suggest that the ACT Benchmarks are effective at

identifying students who are ready for college and likely to succeed beyond the first year of

college.

As a reasonable extension to our earlier study (Radunzel & Noble, 2012a), in this study

we examine the effects of using ACTC score, HSGPA, and the ACT Benchmarks for predicting

college success among student demographic groups. When the relationships between college

outcomes, test scores, and HSGPAs differ among various population student groups, using a

total-group prediction equation (as would be the case in the college admissions process) may

result in systematic over- or underprediction for different student groups (i.e., differential

prediction).

Several studies have examined the differential effects on race/ethnicity and/or gender of

using standardized test scores (including ACTC score) and HSGPA to predict first-year college

GPA, thereby helping to ensure equity in the admissions process (Sanchez, 2013; Mattern,

Patterson, Shaw, Kobrin, & Barbuti, 2008; Noble, 2003; Young, 2001). Sanchez (2013) also

examined the differential effects on family income groups of estimating students’ chances of

earning a 2.5 or higher or a 3.0 or higher first-year college GPA based on their ACTC scores

and/or HSGPAs. Results from this latter study suggested that students’ chances of success

estimated from total-group models (all students irrespective of their demographic characteristics)

were overestimated for African American, Hispanic, lower-income, and male students, and were

slightly underestimated for White, higher-income, and female students. These findings held for

both pre-college measures, although HSGPA models generally resulted in greater over- and

underprediction of first-year success by racial/ethnic and family income groups than ACTC score

models did.

In terms of prediction accuracy, ACTC score and HSGPA were somewhat more accurate

predictors of first-year success for African American and Hispanic students than for White

students using the 3.0 or higher first-year GPA criterion. For the 2.5 or higher GPA criterion, the

percentages of correct classifications at optimal total-group selection values (values that

maximized prediction accuracy for the total group of students) were more comparable across

racial/ethnic groups. This latter finding also held for the family income and gender group

comparisons at both GPA criterion levels. Results from the study by Sanchez (2013) are

consistent with earlier studies (Mattern et al., 2008; Noble, 2003; Young, 2001), and suggest that

African American, Hispanic, and lower-income students are not disadvantaged in the college

admission process when ACTC score is used to predict first-year GPA.

Therefore, in this study, we extend the research by Sanchez (2013) to include college

outcomes beyond the first year through degree completion. In particular, in this study we

investigate the differential effects on student demographic groups of using one of the following

sets of pre-enrollment achievement measures to predict college success through degree

completion:

 ACTC score,

 HSGPA,

 ACTC score and HSGPA, or

 the ACT College Readiness Benchmarks.

Using total-group and group-specific predictions based on ACTC score and/or HSGPA, as well

as total-group ACTC score or HSGPA selection values that maximized prediction accuracy, we

compare the probabilities of success and percentages of correct classifications across student

demographic groups and predictor variables. The percentages of successful students for those at

or above the ACT Benchmark scores are also compared among student groups.

Clearly, a student’s likelihood of being successful in college is based on multiple factors,

including both cognitive and noncognitive characteristics (Allen & Robbins, 2010). ACT does

not advocate making college success predictions solely on the basis of a single measure, such as

a test score, or a single selection value. The use here of one or two predictors is a mathematical

simplification that can be generalized to multiple measures.

Data

Data for this study included approximately 194,000 ACT-tested students who enrolled in

college as first-time entering students in fall, 2000 through 2006. Over 100 institutions were

represented, including all public institutions from two state systems. Four-year institutions were

required to have at least six years of follow-up data available on their students. Two-year

institutions were required to have at least three years of follow-up data available on their

students. Multiple freshman cohorts of students from an institution were combined together in

the analyses. Cohort years spanned from 2000 to 2003 for 61 four-year study institutions and

from 2000 to 2006 for 43 two-year institutions. However, some institutions provided data for

some but not all of the outcomes. As a result, the number of institutions and enrolled students

with available data differed by college outcome. For additional information, see Radunzel and

Noble (2012a).

To examine the differential effects on student demographic groups of using ACTC score,

HSGPA, and ACT Benchmark scores to inform college admission decisions, we also included

over 505,000 students who sent their ACT scores to the same 104 institutions during the same

time frame but did not enroll there.

Nonenrolled students were identified from the 2000 to 2006

ACT records of all ACT-tested high school graduates nationally. These students requested that

their ACT test scores be sent to at least one of the 104 institutions included in this study during

the same time period as that for enrolled students. Nonenrolled students who sent scores to an

institution, plus those who actually enrolled in an institution, comprised the “applicant pool” for

that institution. The applicant pools for the institutions in this study were intended to

approximate actual applicant pools.

College outcomes included annual progress to degree (based on cumulative hours

earned), degree completion, and cumulative GPA at 150% of normal time to degree completion

(at the end of year 6 for four-year institutions and the end of year 3 for two-year institutions).

Four-year institutions make admission decisions about applicants. And, although most two-year institutions have

open admission policies, they are still concerned about the level of academic preparedness of their future incoming

students and often work with potential applicants through activities like high school outreach and bridge programs

(Barnett, Corrin, Nakanishi, Bork, Mitchell, Sepanik, … Clabaugh, 2012; Kerrigan & Slater, 2010). An example of a

high school outreach program includes an early assessment/intervention program where two-year institutions offer

high school juniors and seniors the opportunity to take college placement tests to evaluate their level of college

readiness and then encourage them to strengthen and refresh their skills if needed.

Students may send their ACT scores to any number of institutions, but actually apply to only a subset of them.

Conversely, some students may apply to some institutions without submitting official ACT score reports.

Analyses were done separately by institution type, where type was determined at the time of

initial enrollment. Progress to degree outcomes over time approximated bachelor’s degree

completion in about five years for students who started at four-year institutions and associate’s

degree completion in slightly over three years for students who started at two-year institutions;

approximations were based on using thresholds for cumulative hours earned that increased by 24

and 18 hours, respectively, each year. For degree completion, we evaluated earning a bachelor’s

degree within six years of initial enrollment at a four-year institution and earning an associate’s

degree within three years of initial enrollment at a two-year institution. For two-year institutions

from two state systems, we also evaluated associate’s degree completion or transfer to an in-state

four-year institution within three years of initially enrolling in college. Cumulative GPA was

evaluated at the end of year 6 for four-year institutions and at the end of year 3 for two-year

institutions (referred to in this report as the year 6/year 3 cumulative GPA) for enrolled students

and at the time of degree completion for students who graduated with a bachelor’s/associate’s

degree before the end of year 6/year 3. Year 6/year 3 cumulative GPA was evaluated at the

following levels: 3.00 or higher and 3.50 or higher.

The pre-enrollment measures used in this study included ACTC score, HSGPA, and the

ACT College Readiness Benchmarks. The ACT Composite score is the rounded arithmetic

average of the four subject area scores (English, Mathematics, Reading, and Science). Test

scores are reported on a scale of 1 to 36. HSGPA was based on student’s self-reported

coursework taken in up to 23 specific courses in English, mathematics, social studies, and

science and the self-reported grades earned in these courses. The ACT College Readiness

Benchmarks correspond to scores of 18, 22, 21, and 24 on the ACT English, Mathematics,

We are using higher criterion levels for year 6/year 3 cumulative GPA than Sanchez used in his study (2013) for

first-year college GPA because the typical average value across institutions was higher for the later outcome than for

the earlier one (3.1 for year 6 GPA and 2.8 for year 3 GPA vs. 2.7 for first-year GPA).

Reading, and Science tests, respectively (Allen & Sconing, 2005). Students who meet the ACT

Benchmark score have approximately a 50% chance of earning a B or better and approximately a

75% chance of earning a C or better in the corresponding college course or courses (ACT,

2010b).

Differential prediction of college outcomes was evaluated by race/ethnicity, family

income range, and gender. These demographic characteristics were reported by the students at

the time that they registered for the ACT test. For race/ethnicity, underrepresented minority

students (African American, Hispanic, and American Indian/Alaskan Native students combined)

were compared to White students.

In this report, underrepresented minority students are referred

to as minority students. Family income was categorized as less than $30,000 (Low), $30,000 to

$60,000 (Mid), and more than $60,000 (High).

Table 1 provides the typical student

demographic percentages across postsecondary institutions.

The racial/ethnic minority group includes students who are generally underrepresented in postsecondary education.

Results for these racial/ethnic groups were combined to have sufficient sample sizes of underrepresented minority

students included in each institution’s applicant pool (10 or more). At the time of data collection, Native

Hawaiian/other Pacific Islander students (another racial/ethnic group often underrepresented in postsecondary

education) was not a separate racial/ethnic category. Therefore, students of this race/ethnicity could not be included

in the underrepresented minority group in this study. Results for other racial/ethnic groups such as Asian American

students are not reported due to smaller sample sizes.

The US median household income in 2003 was approximately $43,000 (US Census Bureau, 2003).

Table 1

Distributions of Student Demographic Percentages for Institutional Applicant Pools

by Type of Institution

Student demographic

characteristic

Two-year institutions Four-year institutions

Med Min Max Med Min Max

Race/ethnicity

Minority 21 3 51 17 3 93

White 73 46 95 78 4 95

Family income range

Low 39 18 54 25 14 62

Mid 43 34 55 43 28 53

High 17 11 39 29 10 50

Gender

Female 57 32 69 56 41 79

Male 42 33 64 43 21 56

Note. Median family income and gender percentages may not sum to 100 percent due to rounding. Median

racial/ethnic percentages do not sum to 100 percent due to other racial/ethnic student groups not included in

racial/ethnic comparisons, (e.g., Asian American students). Med = median; Min = minimum; Max = maximum.

Method

For each student demographic group and institution, we computed mean ACTC scores

and HSGPAs for enrolled students and the entire applicant pool. Mean cumulative GPAs and

college success rates were also calculated by institution and student demographic group for

enrolled students. Distributions of the means and rates associated with these variables were then

summarized across institutions and student demographic groups using minimum, median, and

maximum values.

We used hierarchical logistic models to estimate progress to degree, cumulative GPA,

and degree completion rates for enrolled students from the pre-enrollment measures and student

demographic characteristics (referred to in this report as group-specific regression models).

The hierarchical logistic regression models were estimated in HLM 6.08 (Raudenbush, Bryk, Cheong, & Congdon,

2004) using the Laplace approximation method. The pre-enrollment achievement measures were included in the

models in their original units (i.e., variables were not centered in the models).

Hierarchical models account for students clustered within institutions and allow the estimated

college success rates to vary across institutions. The pre-enrollment measures of ACTC score

and HSGPA were evaluated individually, as well as jointly, in the models. The individual ACT

subject area scores were each evaluated in separate models. The group-specific models not only

included the pre-enrollment measures and the individual student demographic characteristics but

also interactions between the pre-enrollment measures and student characteristics. We developed

separate models by year of enrollment for each relevant outcome and by institution type (two- vs.

four-year). The intercepts and the slopes of the main effects were included as random effects in

all models; interaction terms were included as fixed effects.

To examine the differential effects of ACTC score or HSGPA on long-term college

success by student demographic group, we used three different approaches. First, to evaluate

differential prediction by student demographic group across the entire ACTC score and HSGPA

scales, we compared typical probabilities of success estimated from the group-specific regression

models to those estimated from the total-group regression models. The total-group regression

models included the pre-enrollment achievement measures only, and did not include any of the

student demographic indicator(s) (described in detail in Radunzel and Noble (2012a)). For both

models, the probabilities of success were derived using the fixed effects parameter estimates

from the models. When the differences in the probability estimates between the total-group and

group-specific models at the same ACTC score or HSGPA are positive, then the total-group

model overpredicts the probabilities of success for the specific student group. When these

differences are negative, then the total-group model underpredicts success for the specific

student group.

Second, we also evaluated differential prediction by student demographic group at total-

group optimal selection values (values that were used to model the use of ACTC scores and

HSGPA for college admissions).

Optimal total-group selection values correspond to a 0.50

probability of success for a given model and maximize the estimated percentages of correct

selection decisions (Sawyer, 1996). Optimal selection values could be determined only for those

institutions whose total-group probability curves crossed 0.50 (that is, institutions with “viable”

models).

For the two-predictor models, multiple combinations of ACTC score and HSGPA

corresponding to a probability of success of 0.50 were identified. The total-group optimal

selection value(s) were used in this study for comparative purposes only (see Radunzel & Noble,

2012a).

In general, institutions rarely use strict selection values and often use multiple measures

in making their admission decisions (Clinedinst, Hurley, & Hawkins, 2011).

For each institution with a viable total-group model, we applied the institution-specific

total-group optimal selection value(s) to the corresponding group-specific probability

distributions for each institution, student demographic group, and predictor (or predictor

combination). We then summarized the distributions of these group-specific probabilities of

success across institutions using minimum, median, and maximum values. A typical (median)

Unlike the first approach, differential prediction is compared at ACTC score and/or HSGPA values that may differ

across institutions, since total-group optimal selection values for ACTC score and/or HSGPA (individually and

jointly) from the total-group regression models were identified for each institution (Radunzel & Noble, 2012a).

Outcomes that resulted in smaller numbers of institutions with viable total-group models included associate’s

degree completion and achieving higher levels of year 6/year 3 cumulative GPA when they were modeled as a

function of HSGPA. The reason for this is that for many institutions, students’ chances of success for these

outcomes were relatively low in general and never reached 50% across the entire HSGPA scale (see Appendix B

from Radunzel & Noble (2012a)).

The typical ACTC or HSGPA values that maximized prediction accuracy (that is, the values associated with at

least a 50% chance of being successful) were relatively high for degree completion from the same initial institution.

However, the optimal selection values also varied substantially across institutions (lower selection values were

generally seen for institutions with higher degree completion rates). In part, typical selection values were so high

because degree completion rates from the same institution were generally low, especially for two-year institutions.

Institutions are able to compensate for lower admissions standards with effective support programs and

interventions. For additional discussion on these matters, see pp. 45-47 from our earlier study (Radunzel & Noble,

2012a).

group-specific probability of success below 0.50 suggests that the total-group model tends to

overpredict success at the total-group optimal selection value(s) for the specific student group,

and a typical probability of success above 0.50 suggests underprediction for the student group.

We also used this approach to evaluate whether over- or underprediction for a particular student

group is consistently observed across all institutions. In the results section, we show that these

first two approaches lead to the same general differential prediction conclusions among the

student demographic groups.

Third, to evaluate the differences in prediction accuracy by student demographic group,

we estimated the following statistics for each predictor/predictor combination and outcome at

institution-specific total-group optimal selection value(s):

1. the percentage of correct classifications (accuracy rate (AR)),

2. the percentage of successful students among those expected to be successful (success

rate (SR)),

3. the increase in the percentage of correct classifications over expecting all applicants

to be successful (increase in accuracy rate (∆AR)), and

4. the percentage of students with values below the selection value(s) (100 minus this

percentage gives the percentage of students in the applicant pool at or above the

selection value(s)).

We calculated these statistics using the institution-specific parameter estimates from the

group-specific regression models and the corresponding group distributions of ACTC scores and

HSGPA for each institution’s applicant pool.

Correct classifications include students at or

above the total-group selection value(s) who would be successful and students below the value(s)

For each institution the estimated group-specific conditional probabilities of success for nonenrolled students were

assumed to be the same as those for enrolled students.

who would have not been successful. For a more complete description of the methodology used

(including the assumptions being made) to evaluate the usefulness of pre-enrollment measures in

the admissions process, see Sawyer (2010).

Distributions of these statistics were summarized across institutions and student groups

using minimum, median, and maximum values. In this paper, we present results across

institutions with viable models for each individual predictor/outcome combination. However,

results across institutions with viable models for both predictors were similar to those reported

here.

To study the differential effects on student demographic groups of using the ACT College

Readiness Benchmarks for predicting college success, we estimated group-specific probabilities of

success and SRs at the Benchmark scores for each institution. Increases in SRs (denoted as SRs)

were also estimated to evaluate the usefulness of the predictor variables for increasing SRs over

baseline success rates. For each student group, we summarized the distributions of probabilities of

success, SRs, and SRs across institutions using minimum, median, and maximum values. To

evaluate the differential prediction of using the ACT Benchmark scores by student group, we

compared typical total-group probabilities of success at the ACT Benchmark scores to

corresponding group-specific probabilities of success (positive differences suggest overprediction

and negative differences suggest underprediction).

In addition, we compared typical values of SRs

and SRs among student groups.

When students completed the ACT registration materials, some of them omitted

responses to high school coursework and grade items, as well as to the family income range item.

Since our focus was on evaluating the specific ACT Benchmark scores, we did not compare the typical

probabilities of success estimated from the total-group and group-specific models (using the fixed-effects parameter

estimates) across the entire scale of possible ACT subject area test scores. Such results at the ACT Benchmark

scores are expected to be comparable to those reported here based on the median value across institutions.

We used multiple imputation to estimate missing values; 12% and 17% of enrolled students and

11% and 15% of nonenrolled students had missing HSGPA and family income range,

respectively. Five data sets were imputed. We developed models for all five imputed data sets.

No practically significant differences in parameter estimates (including standard errors) were

found across the data sets. For all analyses involving HSGPA and family income range we report

the results based on the initial imputed data set.

Results

Differential Effects of ACTC Score and HSGPA for Predicting Long-Term College Success

In this section, we describe the differential effects on student demographic groups of

using ACTC score and HSGPA separately and jointly for predicting college success through

degree completion. We first present descriptive statistics for ACTC scores, HSGPAs, and college

outcomes over time disaggregated by race/ethnicity, family income, and gender. Next, we

present group-specific probability distributions for the various college outcomes as functions of

ACTC scores and HSGPAs, and compare these estimates to those derived from the total-group

models. Following this, we present for each student demographic group the median probabilities

of success, ARs, ∆ARs, and SRs at the total-group optimal ACTC and HSGPA selection values

to evaluate the accuracy of these pre-college measures for informing students’ chances of long-

term college success.

Descriptive statistics. Mean ACTC scores and HSGPAs were typically higher among

enrolled students than among students in the entire applicant pool at four-year institutions, but

means were comparable between enrolled students and the entire applicant pool at two-year

institutions. These findings held when examined by race/ethnicity, family income range, and

gender (Appendix A, Tables A-1 to A-6). For both the enrolled and applicant pool samples,

White students, higher-income students, and female students typically had higher mean ACTC

scores and HSGPA values than minority students, lower-income students, and male students,

respectively, at both two- and four-year institutions.

For most student demographic groups, the typical mean ACTC scores of enrolled

students in this study were lower than mean ACTC scores of first-year ACT-tested college

students nationally who enrolled in college in 2003 (Table A-7). This finding was observed at

both two- and four-year institutions, and is consistent with our previously reported results for the

total group of students (Radunzel & Noble, 2012a). For minority students and lower-income

students (the two exceptions to the general finding for most groups), typical ACTC means were

similar to or slightly higher than the corresponding national means. Differences in mean ACTC

scores between enrolled students nationally and the sample of enrolled students for this study

were larger for male students and higher-income students than for female students and middle-

income students, respectively.

College success rates, including degree completion rates, were typically higher for White

students than for minority students, and higher for female students than for male students (Tables

A-1 and A-2 for race/ethnicity and Tables A-5 and A-6 for gender). For example, the typical six-

year bachelor’s degree completion rate across four-year institutions was 14 percentage points

higher for White students than for minority students (44% vs. 30%) and nearly 10 percentage

points higher for female students than for male students (46% vs. 37%). As family income range

increased, typical college success rates also increased (Tables A-3 and A-4). For example, at

four-year institutions, we found that the typical six-year bachelor’s degree completion rate was

14 percentage points higher for higher-income students than for lower-income students (47% vs.

33%).

The result among income groups held for four-year institutions only.

The same general conclusions by student demographic group were seen at two-year

institutions, albeit with a few exceptions. First, the typical three-year degree completion or

transfer rate was the same for male and female students (23%; Table A-6). Second, the typical

three-year associate’s degree completion rate by income group was highest for middle-income

students, followed by higher-income students (Table A-4). Given that the three-year degree

completion or transfer rate was highest for higher-income students, a possible explanation for the

degree completion result (without transfer) is that higher-income students were more likely to

bypass earning an associate’s degree before transferring to a four-year institution.

Probabilities of success by student demographic group. In Appendix B we provide

figures of the estimated probabilities of completing a degree or achieving levels of year 6/year 3

cumulative GPA as a function of ACTC score or HSGPA by student demographic group

(Figures B-1 to B-18).

We estimated the probabilities in the figures using the fixed effects

parameter estimates from the group-specific hierarchical logistic models. Across college

outcomes and student demographic group, we found that as ACTC score or HSGPA increased,

the estimated probabilities of success at either a typical two- or four-year institution also

increased.

Race/ethnicity. For students with ACTC scores of 27 or below, probabilities of

completing a bachelor’s degree by year 6 for minority students were lower than those for White

students. In comparison, for students with ACTC scores of 28 or above, corresponding

probabilities for minority students were comparable to or higher than those for White students

(Figure B-1). We found a similar result for each of the other outcomes: however, the ACTC

score associated with the change in the direction of the racial/ethnic differences (from negative to

Probabilities of success are shown for degree completion and achieving levels of year 6/year 3 cumulative GPA;

they are not shown for the progress to degree outcomes. In addition, probabilities of college success are shown over

the range of observable ACTC scores and HSGPAs for each student demographic group.

positive) depended on the outcome (Figures B-1, B-3, and B-5). In contrast, probabilities of

college success predicted from HSGPA were consistently lower for minority students than for

White students (Figures B-2, B-4, and B-6). And, unlike the results for ACTC score, we found

that the racial/ethnic differences increased as HSGPA increased. These findings held for all

outcomes at both two- and four-year institutions.

Total-group probabilities estimated from ACTC score or HSGPA individually were

generally similar to or slightly lower than the corresponding group-specific estimates for White

students: The total-group model slightly underpredicted probabilities of success for White

students relative to group-specific probabilities (by, at most, 0.04 across outcomes at both two-

and four-year institutions; Tables C-1 and C-2 in Appendix C). We illustrate this finding for

bachelor’s degree completion by year 6 by ACTC score in Figure 1 and by HSGPA in Figure 2.

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

10 15 20 25 30 35

ACTC score

Probability

White students Minority students Total group

Figure 1. Estimated probabilities of six-year bachelor’s degree completion by ACTC score and

race/ethnicity. ACTC = ACT Composite.

Probabilities estimated from the fixed effects parameter estimates from the total-group hierarchical logistic

models are provided for comparison.

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

2.0 2.5 3.0 3.5 4.0

HSGPA

Probability

White students Minority students Total group

Figure 2. Estimated probabilities of six-year bachelor’s degree completion by HSGPA and

race/ethnicity. HSGPA = high school grade point average.

In comparison, total-group models tended to overpredict probabilities of success for

minority students with ACTC scores at or below the 75th percentile for enrolled students (values

of 25 and 21 at four- and two-year institutions, respectively; Figure 1 and Tables C-1 and C-2).

At four-year institutions, as ACTC score increased beyond the 75

percentile, differences

between total-group probabilities and those for minority students tended to decrease, suggesting

little to no differential prediction of college success by race/ethnicity for students with higher

ACTC scores. This finding was also observed for achieving a year 3 cumulative GPA of 3.00 or

higher, or 3.50 or higher, at two-year institutions (Figure B-5). For the progress to degree and

degree completion outcomes at two-year institutions, results suggested slight underprediction for

the total-group model in estimating probabilities of success for minority students with higher

ACTC scores (at the 99th percentile of ACTC scores of 28 or above; by, at most, 0.06; Table C-

2).

Unlike the results for ACTC score, the amount of overprediction for minority students

increased as HSGPA increased (Figure 2, Tables C-1 and C-2). This finding held for most

outcomes at both two- and four-year institutions. We also found that for both predictors the

largest differences between the total-group and group-specific probabilities for minority students

generally occurred for the outcome of achieving levels of year 6/year 3 cumulative GPA.

Family income. For most of the progress to degree and degree completion outcomes,

probabilities of success were greater for higher-income students than for lower-income

students.

This result held for both ACTC score and HSGPA (Figures B-7 and B-8 for degree

completion). Differences in probabilities of success between higher- and lower-income students

decreased as ACTC score increased (especially at two-year institutions), but the opposite was

true for HSGPA.

For both GPA levels at any given ACTC score, group-specific probabilities of achieving

levels of year 6/year 3 cumulative GPA were generally comparable across family income groups

(Figures B-9 and B-11). In contrast, income group differences in corresponding probabilities of

success associated with HSGPA tended to increase as HSGPA increased at both two- and four-

year institutions, especially for the 3.50 or higher criterion (Figures B-10 and B-12).

Total-group probabilities estimated from either ACTC score or HSGPA were generally

similar to or slightly lower than those for middle-income students. We illustrate this finding for

bachelor’s degree completion by year 6 across the ACTC score scale in Figure 3 and across the

HSGPA scale in Figure 4.

The exception to this finding was for associate’s degree completion by year 3 at two-year institutions. For students

with higher ACTC scores (at the 99th percentile of 28 or above), the chances of completing an associate’s degree by

year 3 were greater for lower-income students than for higher-income students.

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

10 15 20 25 30 35

ACTC score

Probability

High Mid Low Total group

Figure 3. Estimated probabilities of six-year bachelor’s degree completion by ACTC score and

family income. ACTC = ACT Composite.

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

2.0 2.5 3.0 3.5 4.0

HSGPA

Probability

High Mid Low Total group

Figure 4. Estimated probabilities of six-year bachelor’s degree completion by HSGPA and

family income. HSGPA = high school grade point average.

For the progress to degree and degree completion outcomes, total-group models tended to

slightly overpredict probabilities of success for lower-income students and underpredict those

for higher-income students (Figures 3 and 4; Tables C-3 and C-4).

We saw these results at both

types of institutions and for both predictors (by, at most, 0.07 for ACTC score and 0.08 for

HSGPA). For ACTC scores beyond the 75th percentiles, the amount of overprediction and

underprediction for lower- and higher-income students decreased as ACTC score increased, and

approached 0 for higher-income students. In comparison, differences in probabilities resulting

from the total-group and group-specific HSGPA models generally did not decrease for students

with higher HSGPAs. In some cases, the opposite occurred, especially at two-year institutions

(Tables C-3 and C-4).

For year 6/year 3 cumulative GPAs of 3.50 or higher, there was evidence of differential

prediction for lower- and higher-income students using the total-group HSGPA model, but only

for HSGPAs above 3.50 (over- and underprediction by, at most, 0.08 and 0.04, respectively;

Tables C-3 and C-4). A similar result held for the 3.00 or higher criterion at four-year institutions

(by, at most, 0.05 and 0.04, respectively). In contrast, there was minimal differential prediction

by family income group for the 3.00 or higher criterion at two-year institutions using HSGPA.

For both success levels and types of institutions, there was minimal differential prediction by

family income group using the total-group ACTC score model.

Gender. For all outcomes at two- and four-year institutions, probabilities of success

estimated from the ACTC group-specific models were higher for female students than for male

students (Figures B-13, B-15, and B-17). This finding also held for the HSGPA group-specific

models for achieving a 3.00 or higher or 3.50 or higher year 6/year 3 cumulative GPA (Figures

The exception to this result was for the total-group ACTC score model that estimated probabilities for completing

an associate’s degree by year 3. For this outcome, there was slight overprediction for higher-income students with

ACTC scores of 28 or above, and no evidence of overprediction for lower-income students.

B-16 and B-18). Gender differences in probabilities of achieving levels of year 6/year 3

cumulative GPA were generally greater when they were based on ACTC score than when they

were based on HSGPA. For the progress to degree and degree completion outcomes, male and

female students’ chances of success based on HSGPA were comparable (generally within 4

percentage points; Figure B-14 for degree completion).

For all outcomes at both types of institutions, total-group models based on ACTC score

generally overpredicted probabilities of success for male students and, to a lesser extent,

underpredicted those for female students (Figure 5; Tables C-5 and C-6). Across the ACTC score

scale, the maximum amount of over- and underprediction for male and female students was

greater for year 6/year 3 cumulative GPA than for the progress to degree and degree completion

outcomes (0.13 and 0.10 compared to 0.06 and 0.05, respectively).

For all outcomes at four-

year institutions and for year 3 cumulative GPA at two-year institutions, the absolute differences

in probabilities based on the total-group and group-specific models usually decreased as ACTC

score increased for ACTC scores at or above the 75th percentile (Tables C-5 and C-6).

The exception to the progress to degree and degree completion range was for associate’s degree completion by

year 3 where the maximum overprediction for male students was 0.10 and the maximum underprediction for female

students was 0.07.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

10 15 20 25 30 35

ACTC score

Probability

Female Male Total group

Figure 5. Estimated probabilities of six-year bachelor’s degree completion by ACTC score and

gender. ACTC = ACT Composite.

For the progress to degree and degree completion outcomes, total-group probabilities

based on HSGPA were generally within 0.03 of the corresponding group-specific probabilities

for both male and female students (Figure 6 and Tables C-5 and C-6). For cumulative GPA there

was evidence of differential prediction by gender for the total-group HSGPA model

(overprediction for male students by, at most, 0.09 and underprediction for female students by, at

most, 0.06).

Moreover, as HSGPA increased the amount of overprediction for male students

also increased.

For cumulative GPA at two-year institutions, there was minimal underprediction of success for female students

(by, at most, 0.03).

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

2.0 2.5 3.0 3.5 4.0

HSGPA

Probability

Female Male Total group

Figure 6. Estimated probabilities of six-year bachelor’s degree completion by HSGPA and

gender. HSGPA = high school grade point average.

Accuracy and success rates for ACTC score and HSGPA by student demographic

group. In this section, we summarize median probabilities of success, ARs, ∆ARs, and SRs by

student demographic group across institutions with viable models based on ACTC score and/or

HSGPA. These results are evaluated using the institution-specific total-group optimal selection

values (where the total-group probability of success was closest to 0.50) and the group-specific

probabilities of success (Appendix D, Tables D-1 to D-4 for race/ethnicity, Tables D-5 to D-8 for

family income, and Tables D-9 to D-12 for gender).

In our earlier study (Radunzel & Noble, 2012a) for the total group of students, we found

that median ARs, ∆ARs, and SRs for the joint ACTC score and HSGPA models were generally

higher than those based on the single-predictor models. A common finding in this study was that

this result was seen for each student group, regardless of the outcome.

Race/ethnicity. For all three ACTC score and HSGPA model combinations, the typical

probabilities of success at total-group optimal selection values generally exceeded 0.50 for White

students and were less than 0.50 for minority students.

This finding held for most outcomes,

and was in general agreement with our previously reported differential prediction results by

race/ethnicity.

In particular, we found that there was a general tendency for minority students to

have lower probabilities of success than White students with the same ACTC scores/HSGPAs.

However, the direction and magnitude of the differences in the probabilities of success from 0.50

varied across institutions (Tables D-1 to D-4). For example, probabilities of bachelor’s degree

completion by year 6 at the institution-specific total-group ACTC score selection values ranged

from 0.44 to 0.62 for White students and from 0.31 to 0.58 for minority students across

institutions (Table D-1). In general, probabilities of success for minority students were generally

lower than those for White students with the same HSGPA.

These differences were smaller

when based on ACTC score. The total-group model that included both ACTC score and HSGPA

as predictors generally resulted in the smallest amount of underprediction for White students

(probabilities were closer to 0.50). For minority students, the joint model also resulted in less

overprediction than the HSGPA model did.

The total-group selection values were those ACTC scores and HSGPAs that corresponded to a probability of

success of 0.50 (the point where the ARs were maximized for the total group of students).

For outcomes with higher total-group optimal ACTC selection values, we found that there was a tendency for

slight underprediction for minority students (see results for associate’s degree completion with or without transfer by

year 3, Appendix D, Table D-2).

Those that were based on comparing total-group and group-specific probabilities estimated using fixed-effects

parameter estimates from the hierarchical logistic models.

The typical amount of overprediction for minority students and underprediction for White students at total-group

selection values were comparable to those estimated from the fixed-effects models at the same ACTC score or

HSGPA; these estimates were not exactly the same due to differences in the approaches used to combine

information across institutions.

For the later progress to degree outcomes and year 6/year 3 college GPA, we found underprediction for minority

students at the total-group optimal HSGPA selection values for all four-year institutions and for most two-year

institutions included in this study (see minimum and maximum values in Tables D-1 to D-4).

For all three predictor models, median ARs across institutions for most of the long-term

outcomes were higher for minority students than for White students.

Typical increases in

correct classifications (∆ARs) were also substantially higher for minority students than for White

students (e.g., 46% vs. 20%, respectively, for bachelor’s degree completion by year 6 based on

ACTC score). In contrast, typical SRs evaluated at or above the total-group optimal selection

values were generally higher for White students than for minority students (e.g., 56% and 52%

for bachelor’s degree completion based on ACTC score). However, for all outcomes at both two-

and four-year institutions, racial/ethnic differences in median SRs were smaller for the ACTC

score or joint models than for the HSGPA models (e.g., 4, 5, and 10 percentage points,

respectively, for bachelor’s degree completion by year 6). In general, higher percentages of

minority students than White students had ACTC scores and HSGPAs below the institution-

specific total-group optimal selection values. Thus, substantially fewer minority students than

White students had ACTC scores or HSGPAs at or above the total-group optimal values.

Family income. For the progress to degree and degree completion outcomes, the typical

probabilities of success by family income group at the total-group optimal selection values were

generally less than 0.50 for lower-income students (overprediction), near 0.50 for middle-income

students, and above 0.50 for higher-income students (underprediction).

This finding generally

held for all three predictor models for most, if not all institutions (Tables D-5 and D-6), but the

magnitude of over- and underprediction for lower- and higher-income students varied across

institutions. For example, probabilities of bachelor’s degree completion by year 6 estimated from

These outcomes included outcomes beyond year 2 for four-year institutions and beyond year 1 at two-year

institutions.

The exception to this finding was for associate’s degree completion by year 3 at two-year institutions, where

median probabilities of success were comparable across the three income groups.

the ACTC score and HSGPA joint model ranged from 0.40 to 0.49 across institutions for lower-

income students and from 0.53 to 0.58 for higher-income students.

For two-year institutions, typical probabilities of a 3.00 or higher, or 3.50 or higher, year

3 cumulative GPA by income group at total-group optimal selection values were comparable,

irrespective of the predictor used (Table D-8). For the ACTC score and joint models, this result

also held at both GPA criterion levels for year 6 cumulative GPA at four-year institutions (Table

D-7).

Across outcomes using the total-group optimal selection values, we found that HSGPA

tended to overpredict success for lower-income students more than ACTC score, while the extent

of underprediction for higher-income students was more similar for the two predictors.

For the later progress to degree and degree completion outcomes, median ARs across

institutions were typically higher for lower-income students than for middle- and higher-income

students irrespective of predictor (Tables D-5 and D-6). Correspondingly, for most of these

comparisons, median ARs generally decreased as income level increased. We found a similar

result for achieving a year 6/year 3 cumulative GPA of 3.50 or higher (84%, 79%, and 77% for

lower-, middle-, and higher-income students at four-year institutions, Table D-7). For the earlier

progress to degree outcomes and achieving a year 6/year 3 cumulative GPA of 3.00 or higher,

median ARs were more comparable among the income groups. Across outcomes for all three

predictor models, median ∆ARs were substantially greater for lower-income students than for

higher-income students, and generally greater for middle-income students than for higher-

income students (Tables D-5 through D-8).

On the other hand, typical SRs for students at or above the total-group optimal selection

values generally increased as income level increased (e.g., from 49% to 60% for bachelor’s

There was evidence of overprediction for lower-income students and underprediction for higher-income students

at the total-group optimal HSGPA selection values for this outcome at four-year institutions.

degree completion and from 50% to 58% for associate’s degree completion or transfer to a four-

year institution based on ACTC score model).

For most outcomes at two-year institutions,

income group differences in median SRs were generally smaller when ACTC score was used

alone or jointly with HSGPA than when HSGPA was used alone (e.g., 8, 11, and 15 percentage

points, respectively, for associate’s degree completion or transfer to a four-year institution by

year 3; Table D-6). We saw this result for both of the year 6 cumulative GPA outcomes at four-

year institutions (Table D-7). For the progress to degree and degree completion outcomes at

four-year institutions, income group differences in typical SRs were more comparable for all

three predictor models (Table D-5). For each outcome, it was generally the case that higher

percentages of lower-income students than of higher-income students had ACTC scores or

HSGPAs below the institution-specific total-group optimal selection values (e.g., typical

percentages of students below ACTC score selection values for bachelor’s degree completion

decreased from 90% to 78% as income level increased; Table D-5).

Gender. For most of the predictor models and outcomes, the typical probabilities of

success at the total-group optimal selection values were generally above 0.50 for female students

(underprediction) and below 0.50 for male students (overprediction).

Probabilities of success at

the total-group optimal selection values varied across institutions (e.g., from 0.50 to 0.59 for

female students and from 0.41 to 0.52 for male students using ACTC score to predict bachelor’s

degree completion by year 6; Table D-9). In general, using ACTC score alone tended to result in

Exceptions to this finding were for completing an associate’s degree by year 3 and achieving levels of year 3

cumulative GPA. The result for associate’s degree completion by year 3 based on ACTC score model might be

explained by higher-income students with higher ACTC scores being more likely to transfer to a four-year

institution before earning an associate’s degree (as evidenced by larger differences in the probabilities between the

two associate’s degree outcomes for higher-income students than corresponding differences for lower-income

students; Figure B-7).

For the progress to degree and degree completion outcomes at both types of institutions, typical probabilities of

success at the total-group optimal selection values based on the HSGPA or joint models were near 0.50 for both

female and male students. This result suggested minimal differential prediction by gender at the total-group optimal

selection values for these outcomes and predictors.

slightly greater differential prediction by gender at the total-group optimal selection values than

when HSGPA was used alone or in combination with ACTC score (Tables D-9 to D-12).

For all three predictor models and most of the outcomes considered in this study, typical

ARs at total-group optimal selection values were similar for female and male students.

Gender

differences in median increases in correct classifications (∆ARs) were larger at four-year

institutions than at two-year institutions, with median ∆ARs consistently higher for male students

than for female students. Conversely, typical SRs associated with total-group optimal selection

values were consistently higher for female students than for male students (e.g., 61% and 51%

for bachelor’s degree completion using ACTC score). Gender differences in median SRs were

smaller when HSGPA was used alone or jointly with ACTC score than when ACTC score was

used alone (e.g., 5, 5, and 10 percentage points, respectively, for bachelor’s degree completion

by year 6). For each outcome, typical percentages of students scoring below the institution-

specific total-group optimal ACTC score selection values were similar for female and male

students. In contrast, for HSGPA, corresponding median percentages were higher for male

students than for female students.

Differential Effects on Student Demographic Groups of Using ACT College Readiness

Benchmarks for Predicting Long-Term College Success

In this section, we evaluate the differential effects of the ACT College Readiness

Benchmarks for predicting college success through degree completion among student

demographic groups. We first present descriptive statistics on ACT Benchmark attainment for

enrolled students, as well as for the entire applicant pool disaggregated by race/ethnicity, family

For year 6/year 3 cumulative GPA, there was evidence of overprediction for male students and underprediction

for female students at the total-group optimal selection values for HSGPA (by, at most, 0.09 for the typical

probabilities of success; Tables D-11 and D-12).

The exception was for achieving a year 6/year 3 cumulative GPA of 3.50 or higher (Tables D-11 and D-12);

median ARs were slightly higher for male students than for female students.

income, and gender. We then evaluate the typical probabilities of success, SRs, and ∆SRs

associated with the ACT College Readiness Benchmark scores among student demographic

groups.

Descriptive statistics. At four-year institutions, the typical percentages of students

meeting the ACT Benchmarks were higher among enrolled students than among students in the

applicant pool. In contrast, the typical Benchmark attainment percentages for students at two-

year institutions were comparable for the enrolled and applicant pool samples, and were lower

than those for students at four-year institutions. These findings were consistently seen when

disaggregated by race/ethnicity, family income range, and gender (Appendix E, Tables E-1 to E-

3).

At both two- and four-year institutions, median percentages of students meeting the ACT

Benchmarks were substantially higher for White students than for minority students and for

higher-income students than for lower-income students (Tables E-1 and E-2). Typical

Benchmark attainment percentages were slightly higher for female students than for male

students in English and reading, but were slightly lower in mathematics and science (Table E-3).

Probabilities of success and success rates for ACT College Readiness Benchmarks

by student demographic group. For each student demographic group, we calculated median

probabilities of success, SRs, and ∆SRs associated with ACT Benchmark scores across all

institutions with available outcome data. For these analyses, we evaluated year 6/year 3

cumulative GPA for the 3.00 or higher criterion level only.

A common finding that we observed across student demographic groups was that the

typical probabilities of success, SRs, and ∆SRs were generally higher for the ACT Mathematics

For this criterion level, the ACT Benchmark scores are more similar to the typical total-group optimal ACTC

score selection values.

and Science Benchmarks than for the ACT English and Reading Benchmarks. This finding is

consistent with our previously reported result for the total group of students (Radunzel & Noble,

2012a).

Race/ethnicity. The probabilities of success at the Benchmark scores for both White and

minority students varied substantially across institutions. For example, probabilities of bachelor’s

degree completion by year 6 estimated at the ACT English Benchmark score ranged from 0.12 to

0.77 for White students and from 0.10 to 0.64 for minority students. Median probabilities of

success associated with the ACT College Readiness Benchmarks were higher for White students

than for minority students (Appendix F, Tables F-1 and F-2). However, racial/ethnic differences

in the median probabilities of success at the Benchmark scores were smaller than those in the

observed median proportions of success, where prior achievement was not considered (see

Tables A-1 and A-2). For example, for bachelor’s degree completion by year 6, racial/ethnic

differences in probabilities ranged from 0.05 to 0.09 across Benchmark subject areas (e.g., 0.34

vs. 0.28 for the English Benchmark), compared to an observed difference in proportions

(irrespective of Benchmark attainment) between White and minority students of 0.14 (0.44 vs.

0.30). This result generally held across outcomes at both two- and four-year institutions.

For White students, median probabilities of success at the Benchmark scores were similar

to those for the total group of students (generally differed by only 0.01 to 0.02).

For minority

students, probabilities of college success at the Benchmarks were typically lower than those for

For achieving a year 6 cumulative GPA of 3.00 or higher, the typical group-specific probabilities of success at the

ACT English and Reading Benchmark for White students were greater than those estimated from the total-group

model by 0.04 (that is, the total-group model slightly underpredicted the success of White students for this outcome;

Table F-1).

the total group of students.

Thus, the total-group models tended to overpredict college success

for minority students scoring at the Benchmarks. It was generally the case that there was less

overprediction associated with the ACT Mathematics Benchmark than for the other

Benchmarks.

For example, for bachelor’s degree completion by year 6, differences between the

typical probability estimates for the total group and those for minority students were 0.07, 0.04,

0.07, and 0.05 at the ACT English, Mathematics, Reading, and Science Benchmarks,

respectively.

Typical SRs associated with the ACT Benchmark scores were slightly higher for White

students than for minority students. This finding held for all outcomes at both two- and four-year

institutions and for each of the four Benchmarks (Tables F-1 and F-2). For example, for

bachelor’s degree completion by year 6, the median SR associated with the ACT Mathematics

Benchmark was 53% for White students, compared to 48% for minority students. Racial/ethnic

differences in median SRs were generally smaller for the ACT Mathematics Benchmark than for

the other Benchmarks (e.g., 10, 5, 9, and 7 percentage points for the English, Mathematics,

Reading, and Science Benchmarks, respectively, for bachelor’s degree completion by year 6).

In contrast, ∆SRs associated with the Benchmark scores were typically higher for minority

students than for White students.

Family income. Median probabilities of success associated with the ACT Benchmarks

were greater for higher-income students than for middle- and lower-income students; lower-

income students tended to have the lowest estimated probabilities of success (Appendix F,

Group-specific probabilities of success for minority students were typically lower than total-group probabilities at

the Benchmark scores by 0.02 to 0.09 for four-year institutions and by 0.02 to 0.07 for two-year institutions. The

one exception to this result was for achieving a year 6 cumulative GPA of 3.00 or higher at four-year institutions,

where the total-group model typically overpredicted success for minority students by 0.11 to 0.16.

Differences in the amount of overprediction for minority students between Benchmarks were relatively small

(ranged from 0.02 to 0.07 for four-year institutions and from 0.01 to 0.04 for two-year institutions).

Exceptions to this finding were for the two degree completion outcomes at two-year institutions where

racial/ethnic differences in median SRs were more comparable across the Benchmarks (Table F-2).

Tables F-3 and F-4).

However, differences in median probabilities of success across income

groups at the Benchmark scores were somewhat smaller than differences in the observed median

proportions of success reported previously (Tables A-3 and A-4), when prior achievement was

not taken into accounted (by, at most, 0.08). This finding held for most outcomes at both types of

institutions.

Median probabilities of success for middle-income students at the Benchmark scores

were similar to corresponding median total-group probabilities (higher by, at most, 0.03).

Probabilities at the Benchmarks for lower-income students were typically lower than the

corresponding total-group probabilities (by 0.02 to 0.09 at four-year institutions and by 0.01 to

0.06 at two-year institutions).

Conversely, typical probabilities for higher-income students were

greater than the corresponding median total-group probabilities (by 0.01 to 0.06 at four-year

institutions and by 0.03 to 0.07 at two-year institutions).

These results taken together suggest

that across the outcomes, total-group models for predicting students’ chances of college success

at the ACT Benchmark scores tended to slightly overpredict probabilities of success for lower-

income students and underpredict success for higher-income students.

Typical SRs associated with the ACT Benchmark scores generally increased as family

income range increased (e.g., from 38% to 50% for bachelor’s degree completion by year 6 using

the ACT English Benchmark). This finding held for most outcomes at two- and four-year

Exceptions to this finding were for completing an associate’s degree by year 3 and achieving a year 3 cumulative

GPA of 3.00 or higher. For these outcomes, median probabilities of success were comparable across the family

income groups (Table F-4).

The one exception to this finding was for achieving a year 3 cumulative GPA of 3.00 or higher at two-year

institutions. For this outcome, group-specific probabilities for lower-income students were similar to the total-group

probabilities (Table F-4).

Exceptions to this finding were for completing an associate’s degree by year 3 and achieving a year 3 cumulative

GPA of 3.00 or higher. For these outcomes, group-specific probabilities at the Benchmark scores for higher-income

students were more similar to the corresponding total-group probabilities (Table F-4).

institutions and across all four Benchmarks (Tables F-3 and F-4).

Income differences in median

SRs were generally smaller for the ACT Mathematics Benchmark than for the other Benchmarks

(e.g., 12, 9, 13, and 11 percentage points for the English, Mathematics, Reading, and Science

Benchmarks, respectively, for bachelor’s degree completion by year 6). Typical ∆SRs associated

with the Benchmark scores were higher for lower-income students than for higher-income

students.

Gender. Median probabilities of success associated with the ACT Benchmarks were

generally higher for female students than for male students (Appendix F, Tables F-5 and F-6).

Gender differences in median probabilities of success were slightly larger for the ACT

Mathematics and Science Benchmarks than for the English and Reading Benchmarks.

For the progress to degree and degree completion outcomes, median probabilities of

success at the Benchmark scores were slightly higher than those for the total group of students

for female students (underprediction), and lower for male students (overprediction) (generally

only by 0.01 to 0.08). For each of the outcome and Benchmark combinations, the

underprediction for female students generally corresponded to a somewhat similar degree of

overprediction for male students. For achieving a year 6/year 3 cumulative GPA of 3.00 or

higher, there tended to be more differential prediction by gender associated with the Benchmark

scores than was seen for the other outcomes (absolute differences in median probabilities ranged

from 0.04 to 0.14 for this outcome). In addition, the degree of overprediction for male students

was larger than the degree of underprediction for female students. These findings were in general

Exceptions to this finding included completing an associate’s degree by year 3 and achieving a year 3 cumulative

GPA of 3.00 or higher. For these outcomes, median SRs were more comparable across the three family income

groups (Table F-4).

Differences in typical ∆SRs between lower- and higher-income students were generally larger for the ACT

Mathematics and Science Benchmarks than for the English and Reading Benchmarks.

agreement with the ones noted earlier for the probabilities of success evaluated at the total-group

optimal ACTC score selection values.

Typical SRs associated with the ACT Benchmark scores were higher for female students

than for male students. This finding held for all outcomes at two- and four-year institutions and

across all four Benchmarks (Tables F-5 and F-6). Gender differences in median SRs were

slightly larger for the ACT Mathematics and Science Benchmarks than for the English and

Reading Benchmarks (e.g., 12 vs. 7 to 8 percentage points, respectively, for bachelor’s degree

completion by year 6). In addition, typical ∆SRs associated with the ACT Benchmarks were

slightly higher for female students than for male students in mathematics and science,

and were

similar for female and male students in English and reading.

Discussion

In this study, we evaluated the differential effects on racial/ethnic, family income, and

gender groups of using ACTC scores and HSGPAs for identifying possible applicants who are

likely to progress towards and complete a degree. This study is unique in that it is the first study

to evaluate differential prediction and differences in prediction accuracy by student groups for

multiple measures of college success beyond the first year at both two- and four-year institutions.

For the most part, results from this study are in general agreement with those from prior studies

that examined first-year college grades or GPA as the outcome (Sanchez, 2013; Mattern et al.,

2008; Noble, 2003; Young, 2001; results previously summarized in the Introduction section).

In this study, we used multiple approaches to examine the differential effects among

student demographic groups, including: (1) comparing the total-group and group-specific

probabilities of success across the entire ACTC score and HSGPA scales (using the fixed effects

The exception to this finding was for achieving a year 6 cumulative GPA of 3.00 or higher. For this outcome,

median ∆SRs associated with the Benchmarks were similar between male and female students (Table F-5).

parameter estimates for the models), (2) evaluating how much the typical and individual

institutional group-specific probabilities of success at the total-group optimal selection values

differed from 0.50, and (3) comparing the prediction accuracy at the total-group optimal

selection values among the student groups. The first two approaches that examined differential

prediction among student groups led to the same general conclusions across institutions for each

student demographic group (discussed in detail below). The latter two approaches demonstrated

that the degree of over- and underprediction, as well as the percentages of correction

classifications, at the total-group optimal selection values varied across institutions (see

minimum and maximum values in Appendices D and F). The third approach revealed a common

finding that was seen across student demographic groups, namely that using both ACTC score

and HSGPA jointly improved prediction accuracy and success rates for most of the outcomes

over those based on HSGPA alone.

Below, we summarize the general findings for each student demographic group, point out

the results from our study that differ from prior research, and discuss some possible explanations

for the results.

Race/Ethnicity

Results from this study suggest that racial/ethnic minority students are not disadvantaged

when ACT test scores are used to help inform college admissions decisions and to identify those

students who are likely to be successful in college beyond the first year. This statement is

supported by the results showing that ACTC score was somewhat of a more accurate predictor of

long-term college success for racial/ethnic minority students than for White students.

Depending on the outcome, accuracy rates at total-group optimal selection values were 2 to 14 percentage points

higher for underrepresented minority students than for White students. Differences in accuracy rates among

racial/ethnic groups were larger for the later college outcomes and for achieving higher levels of year 6/year 3

cumulative GPA.

addition, for all the outcomes examined in this study, increases in the percentages of correct

classifications associated with using these measures over not using them (i.e., selecting all

students) were substantially greater for minority students than for White students. These findings

were also seen when HSGPA was used as the predictor.

We also found that total-group models based on either ACTC score, HSGPA, or both

ACTC score and HSGPA generally overestimated minority students’ chances of success.

Overprediction of long-term college success for minority students was more pronounced when

HSGPA was used, rather than ACTC score. Furthermore, the degree of overprediction for

minority students generally increased as HSGPA increased, and decreased as ACTC score

increased. Overprediction for minority students was also observed at the ACT Benchmark

scores. Little to no underprediction was found for White students using any of the pre-college

measures.

Other first-year college outcome studies (Sanchez, 2013; Mattern et al., 2008; Noble,

2003; Zwick & Sklar, 2005) evaluated differences among individual racial/ethnic minority

groups. These studies found greater overprediction for African American students than for

Hispanic or White students, a finding also reported in a comprehensive review of earlier studies

(Young, 2001). Sanchez (2013) also found slightly greater prediction accuracy for African

American students than for Hispanic or White students for first-year GPAs of 3.00 or higher.

Unfortunately in this study, due to smaller numbers of minority students, we were unable to

examine results for long-term college success by the individual racial/ethnic minority groups.

A consistent finding across studies is that the overprediction of college success for

racial/ethnic minority students tends to be more severe when HSGPA is used alone, compared to

Total-group predictions based on ACTC score were generally found to overestimate underrepresented minority

students’ likelihood of long-term college success by, at most, 0.11, and to underestimate White students’ likelihood

of success by, at most, 0.04 across outcomes.

using test scores alone or jointly with HSGPA. This finding might be explained by racial/ethnic

differences in academic preparation and/or educational opportunities (e.g., attending

underresourced, understaffed schools and not having access to sufficiently rigorous high school

coursework (ACT, 2010b)). These differences in academic preparation are generally reflected in

standardized test scores (ACT, 2005), but may not be in HSGPA. For example, because of large

disparities between high schools in their grading practices and the rigor of their courses, a high-

ranking or high GPA student from one school could differ substantially from a high-ranking or

high GPA student from another institution in his/her preparedness for college-level work.

HSGPA can also be affected by grade inflation (Woodruff & Ziomek, 2004).

Across the outcomes considered in this study, the typical degree of overprediction for

minority students associated with using standardized test scores or HSGPA alone generally fell

within the wide range of reported values from earlier studies (where first-year GPA was the

outcome). In this study, there was greater overprediction for minority students at four-year

institutions for achieving levels of year 6 cumulative GPA than there was for the progress to

degree and degree completion outcomes. At two-year institutions the degree of overprediction

for racial/ethnic minority students was more comparable across the outcomes, and was similar to

that seen at four-year institutions for the progress to degree outcomes. None of the other studies

that we reviewed reported results by institution type: they either included four-year institutions

only in their sample or combined results for two- and four-year institutions.

As previously suggested in the literature (Zwick & Sklar, 2005; Young, 2001),

differential prediction of college success for racial/ethnic groups might be due to racial/ethnic

differences in other factors (cultural, societal, or institutional) that influence students’ likelihood

of success in college. For example, students in the minority on a campus with little diversity in

the study body composition may experience more feelings of anxiety, be less socially engaged in

college, and have a more difficult time making the transition from high school to college (Carter,

2006). And, as a result, they may perform below expectations based on their pre-college

achievement. Research has also shown that minority students tend to be less knowledgeable

about the steps that are needed to prepare for higher education, such as knowing how to finance a

college education or plan for educational and career goals (Tym, McMillion, Barone, & Webster,

2004), thereby putting them at somewhat of a disadvantage for succeeding in college.

Smaller percentages of minority than White students in the applicant pool were at or

above the total-group optimal selection values or ACT Benchmark scores. Minority students are

typically not as academically prepared for college (as evidenced by lower average ACTC scores

and HSGPAs, and being less likely to meet the ACT College Readiness Benchmarks (ACT,

2012)). However, racial/ethnic differences in chances of long-term college success and in

percentages of successful students at the ACT Benchmark scores were smaller than racial/ethnic

differences in the observed college success rates (irrespective of Benchmark attainment). In

addition, racial/ethnic differences in chances of college success and in percentages of successful

students (from among those expected to be successful) were smaller when examined by ACTC

score than by HSGPA. These findings agree with results from two earlier studies (Radunzel &

Noble, 2012b; ACT, 2010b) that showed that college readiness helps reduce racial/ethnic gaps in

college success rates. For example, one of the studies (Radunzel & Noble, 2012b) found that

differences in six-year bachelor’s degree completion rates between African American or

Hispanic students and White students were reduced by more than 50% among those who had met

all four ACT College Readiness Benchmarks.

Family income

Results from this study suggest that lower-income students are not disadvantaged when

ACT test scores are used to help inform college admissions decisions and identify those students

who are likely to be successful in college beyond the first year. This statement is supported by

the result that prediction accuracy using ACTC score total-group optimal selection values for

lower-income students was greater than or equivalent to that for higher-income students.

addition, for all the outcomes examined in this study, increases in the percentages of correct

classifications associated with using pre-college academic measures over not using them were

greater for lower-income students than for higher-income students. These findings were also

seen when HSGPA was used as the predictor.

In this study, we also found that total-group models based on either ACTC score,

HSGPA, or both predictors slightly overpredicted the chances of lower-income students

progressing towards and completing a degree.

For these same outcomes, students’ chances

were slightly underpredicted for higher-income students; there was little to no differential

prediction for middle-income students. For the year 6/year 3 cumulative GPA outcomes, there

was little to no differential prediction among family income groups based on ACTC score. This

finding did not hold for HSGPA. In fact, the degrees of over- and underprediction for lower- and

higher-income students across outcomes were slightly greater for HSGPA than for ACTC

score.

In addition, there was a tendency for over- and underprediction for lower- and higher-

Depending on the outcome, accuracy rates at total-group optimal selection values were 0 to 10 percentage points

higher for lower-income students than for higher-income students. Differences in accuracy rates among income

groups were larger for the later college outcomes and for achieving higher levels of year 6/year 3 cumulative GPA.

Total-group predictions based on ACTC score were generally found to overestimate lower-income students’

likelihood of long-term college success by, at most, 0.07, and to underestimate higher-income students’ likelihood of

success by, at most, 0.07 across outcomes.

This finding might be explained by differences between income groups in academic preparation and/or

educational opportunities (a reason previously suggested for why there is a larger degree of differential prediction by

race/ethnicity associated with HSGPA than with ACTC score).

income students to increase as HSGPA increased, and decrease or remain the same as ACTC

score increased. Total-group models at the ACT Benchmark scores also slightly overpredicted

students’ chances of progressing towards and completing a degree for lower-income students and

underpredicted them for higher-income students.

Sanchez (2013), the only other published study that evaluated differential prediction for

family income groups, found similar prediction accuracy results for first-year GPA to those

reported here for long-term outcomes. One difference in the ACTC score results between these

two studies was that there was evidence of differential prediction among family income groups

for first-year GPA (Sanchez study) but not for year 6/year 3 cumulative GPA in this study.

However, GPA results for this study were based on a more homogenous sample than the Sanchez

study, and were disaggregated by institution type. In particular, this study was based on students

who were still enrolled six (three) years later or had completed a bachelor’s (associate’s) degree

prior to the end of year 6 (year 3): that is, those students who were persisting and/or succeeding

in college through year 6 (year 3).

Differential prediction in college success by family income might be due to income group

differences in other factors that influence students’ chances of success in college. These might

include those previously discussed for race/ethnicity (cultural, societal, or institutional). But,

lower-income students are also more likely than their peers to be first-generation students and to

have non-academic obligations; they are also more likely to have work and/or family

responsibilities that can influence their study habits and chances of long-term college success

(Hurtado, Laird, & Perorazio, 2010; Engle & Tinto, 2008).

In contrast to that seen for race/ethnicity, income group differences in students’ chances

of progressing towards and completing a degree were slightly but not dramatically reduced when

pre-college achievement was taken into account. Another study (Radunzel & Noble, 2012b) also

found that reductions in family income gaps in retention and degree completion rates were

generally smaller than those found for racial/ethnic groups when the number of ACT

Benchmarks met was taken into account. These findings highlight the different types of obstacles

lower-income students face compared to other student groups, even among those who are

academically prepared for college. It has been suggested in the literature that these obstacles can

be overcome with financial assistance and effective institutional support programs (Hurtado,

Laird, & Perorazio, 2010).

Gender

Results from this study suggest that ACT test scores are useful in helping inform college

admissions decisions and identifying those male and female students who are likely to be

successful in college beyond the first year. Percentages of correct classifications were

comparable for male and female students for almost all outcomes examined. In addition,

increases in prediction accuracy associated with using the pre-college academic measures

(compared to not using them) were relatively large for both gender groups, especially for the

later outcomes and for achieving a year 6/year 3 cumulative GPA of 3.50 or higher.

Consistent with results previously reported on the ACT and SAT test scores (Sanchez,

2013; Mattern et al., 2008; Young, 2001) using first-year college GPA as the outcome, we found

that year 6/year 3 cumulative GPA was underpredicted for female students and overpredicted for

male students.

There was slightly greater under- and overprediction by gender associated with

ACTC score than with HSGPA. In comparison, the degree of differential prediction by gender

was less pronounced for the progress to degree and degree completion outcomes than for the

Total-group predictions based on ACTC score were generally found to overestimate male students’ likelihood of

long-term college success by, at most, 0.13, and to underestimate female students’ likelihood of success by, at most,

0.10.

cumulative GPA outcomes. ACT Benchmark scores also slightly underpredicted longer-term

college outcomes for female students and overpredicted them for male students.

As previously suggested in the literature (Mattern et al., 2008; Young, 2001), a plausible

explanation for female students doing better and male students doing worse than predicted, given

the same ACT test score might be due to differences in noncognitive characteristics between the

two gender groups. Prior research (Robbins, Allen, Casillas, Peterson, & Le, 2006; Allen,

Robbins, Casillas, & Oh, 2008) has shown that academic behaviors provide additional

information that increases accuracy in identifying students who are at risk of poor grades in

college and for dropping out, beyond measures of academic achievement. It has also been shown

that female students score slightly higher on scales of noncognitive skills in the areas of

academic discipline, commitment to college, and study skills (Le, Casillas, Robbins, & Langley,

2005). Other research has suggested that female students are also more likely than male students

to seek out and use support services at postsecondary institutions (Angrist, Lang, & Oreopoulos,

2009), which can help improve students’ study skills and chances of success in college.

A study by Allen & Robbins (2010) found that even after controlling for first-year

academic performance, motivation (as measured by the Academic Discipline scale from

ENGAGE

measured at onset of college), and interest-major congruence, male students were

less likely than female students to complete a bachelor’s degree in a timely manner. This finding

supports the hypothesis that there are gender differences in other characteristics associated with a

student’s likelihood of progressing towards and completing a degree. Later measures of

motivation would better capture the level of motivational skills that emerge in response to the

student’s college environment.

A reasonable explanation as to why the differential prediction by gender seems to be less

severe for HSGPA than for ACT test scores is that HSGPA likely measures aspects of both the

cognitive and noncognitive components of college success. For example, HSGPA is not only

affected by level of content mastery, but is also affected by a student’s personal behaviors, such

as whether the student is prudent about taking good notes, putting forth effort and participating in

class, completing homework assignments, and preparing well for course exams. ACT test scores,

on the other hand, measure only the cognitive components. This interpretation is supported by

results from a prior study (Allen et al., 2008) that found that students’ level of academic self-

discipline was statistically related to HSGPA, but was not related to ACTC score.

Conclusions

Some researchers have suggested that standardized test scores like the ACT test are not

useful and not predictive of long-term college success, especially for underrepresented minority

and lower-income students (Soares, 2012). Other researchers have reported that SAT/ACT test

scores add little information to predicting long-term college success, including degree

completion, after statistically controlling for HSGPA (Bowen, Chingos, & McPherson, 2009).

However, results from this study and from our earlier study (Radunzel & Noble, 2012a) do not

support this view.

Regardless of student demographic group, ACTC score prediction accuracy for progress

to degree and degree completion is moderately high (62% to 73% at four-year institutions and

62% to 86% at two-year institutions). For most of the outcomes and student demographic groups

examined, using both ACTC score and HSGPA improves prediction accuracy and identification

of successful students among those expected to be successful. Moreover, overprediction of long-

term college success for underrepresented minority and lower-income students was more severe

when HSGPA was used alone, compared to when ACTC score was used alone or in combination

with HSGPA in the prediction models. In particular, differences in students’ chances of college

success among racial/ethnic and family income groups were smaller when examined by ACTC

score than by HSGPA. These findings taken together provide further evidence of the incremental

benefit of using both ACTC score and HSGPA for predicting college success beyond the first

year. Using multiple measures is consistent with ACT’s recommended usage for college success

predictions. In addition, the ACT Benchmark scores were found to be useful for predicting long-

term college success for all students, irrespective of student demographic characteristics,

providing further validity evidence for using them as measures of college readiness.

Other studies (Robbins, et al., 2006; Lotkowski, Robbins, & Noeth, 2004) have shown

that students who are academically prepared for college, academically self-disciplined, socially

engaged, and committed to college are more likely than those who are not to persist to degree

completion. These findings together with the results from this study also suggest that there is a

need to ensure that all students are offered guidance and have the opportunity to connect their

educational aspirations to solid academic preparation and behaviors in high school, thereby

better preparing and equipping them to succeed in college or career. Given the findings from this

study for pre-enrollment achievement measures, future research should examine whether the

effects of noncognitive student characteristics on early and long-term college success differ

among student demographic groups after accounting for pre-college academic achievement.

References

ACT. (2005). Courses count: Preparing students for postsecondary success. Iowa City, IA:

Author.

ACT. (2010a). Mind the gaps: How college readiness narrows achievement gaps in college

success. Iowa City, IA: Author.

ACT. (2010b). What are ACT’s college readiness benchmarks? Iowa City, IA: Author.

ACT. (2012). ACT profile report–national: Graduating class 2012–national. Iowa City, IA:

Author.

Allen, J., & Robbins, S. (2010). Effects of interest-major congruence, motivation, and academic

performance on timely degree attainment. Journal of Counseling Psychology, 57(1), 23-

35.

Allen, J., Robbins, S., Casillas, A., & Oh, I. (2008). Third-year college retention and transfer:

Effects of academic performance, motivation, and social connectedness. Research in

Higher Education, 49, 647-664.

Allen, J., & Sconing, J. (2005). Using ACT Assessment scores to set benchmarks for college

readiness. (ACT Research Report No. 2005-3). Iowa City, IA: ACT, Inc.

Angrist, J., Lang, D., & Oreopoulos, P. (2009). Incentives and services for college achievement:

Evidence from a randomized trial. American Economic Journal: Applied Economics,

1(1), 1-28.

Barnett, E.A., Corrin, W., Nakanishi, A., Bork, R.H., Mitchell, C., Sepanik, S., … Clabaugh, N.

(2012). Preparing high school students for college: An exploratory study of college

readiness partnership programs in Texas. Retrieved February 5, 2013 from

http://www.postsecondaryresearch.org/i/a/document/22458_NCPR_CRPFullReport_051

712.pdf.

Bowen, W.G., Chingos, M.M., & McPherson, M.S. (2009). Crossing the finish line: Completing

college at America’s public universities. Princeton, NJ: Princeton University Press.

Breland, H., Maxey, J., Gernand, R., Cumming, T., & Trapani, C. (2002). Trends in college

admission 2000: A report of a survey of undergraduate admissions policies, practices,

and procedures. Retrieved May 10, 2012 from the Association for Institutional Research

website: http://www.airweb.org/images/trendsreport.pdf.

Carter, D.F. (2006). Key issues in the persistence of underrepresented minority students. New

Directions for Institutional Research, 130, 33-46.

Clinedinst, M.E., Hurley, S.F., & Hawkins, D.A. (2011). 2011 State of College Admission.

Alexandria, VA: National Association for College Admissions Counseling.

Engle, J., & Tinto, V. (2008). Moving beyond access: College success for low-income, first-

generation students. Washington, DC: The Pell Institute.

González, J. (2012, April 22). Education for all? 2-year colleges struggle to preserve their

mission. The Chronicle of Higher Education. Retrieved on May 10, 2012 from

http://chronicle.com/article/2-Year-Colleges-Fight-to-Save/131608/.

Habley, W., Valiga, M., McClanahan, R., & Burkum, K. (2010). What works in student

retention? Fourth National Survey: Community Colleges Report. Iowa City, IA: ACT,

Inc.

Higher Education Research Institute. (2011, November). Completing college: Assessing

graduation rates at four-year institutions. HERI Research Brief, UCLA, Los Angeles,

CA: HERI.

Hurtado, S., Laird, T.F.N., & Perorazio, T.E. (2010). The transition to college for low-income

students: The impact of the Gates Millennium Scholars Program. Retrieved October 23,

2012 from http://www.gatesfoundation.org/learning/Documents/

Final-TransitiontoCollege-Hurtado.pdf.

Kerrigan, M.R., & Slater, D. (2010). Collaborating to create change: How El Paso community

college improved the readiness of its incoming students through Achieving the Dream

(Culture of Evidence Series, Report No. 4). New York, NY: Columbia University,

Teachers College, Community College Research Center.

Le, H., Casillas, A., Robbins, S.B., & Langley R. (2005). Motivational and skills, social, and

self-management predictors of college outcomes: Constructing the student readiness

inventory. Educational and Psychological Measurement, 65(3): 482-508.

Lotkowski, V.A., Robbins, S.B., & Noeth, R.J. (2004). The role of academic and non-academic

factors in improving college retention. Iowa City, IA: ACT, Inc.

Mattern, K.D., Patterson, B.F., Shaw, E.J., Kobrin, J.L., & Barbuti, S.M. (2008). Differential

validity and prediction of the SAT

. (College Board Research Report No. 2008-4). New

York: The College Board.

Moore, C., Shulock, N., & Offenstein, J. (2009). Steps to success: analyzing milestone

achievement to improve community college student outcomes. Sacramento, CA: Institute

for Higher Education Leadership & Policy.

Noble, J. (2003). The effects of using ACT Composite score and high school average on college

admission decisions for racial/ethnic groups. (ACT Research Report No. 2003-1). Iowa

City, IA: ACT, Inc.

Radunzel, J., & Noble, J. (2012a). Predicting long-term college success through degree

completion using ACT

Composite score, ACT Benchmarks, and high school grade point

average. (ACT Research Report No. 2012-5). Iowa City, IA: ACT, Inc.

Radunzel, J., & Noble, J. (2012b). Tracking 2003 ACT

-tested high school graduates: College

readiness, enrollment, and long-term success. (ACT Research Report No. 2012-2). Iowa

City, IA: ACT, Inc.

Raudenbush, S., Bryk, A., Cheong, Y.F., & Congdon, R. (2004). HLM 6: Hierarchical linear

and nonlinear modeling. Lincolnwood, IL: Scientific Software International, Inc.

Robbins, S., Allen, J., Casillas, A., Peterson, C., & Le, H. (2006). Unraveling the differential

effects of motivational and skills, social, and self-management measures from traditional

predictors of college outcomes. Journal of Educational Psychology, 98, 598-616.

Sanchez, E. (2013). Differential effects of using ACT

College Readiness Assessment scores and

high school GPA to predict first-year college GPA among racial/ethnic, gender, and

income groups. (ACT Research Report No. 2013-4). Iowa City, IA: ACT, Inc.

Saupe, J.L., & Curs, B.R. (2008). Deriving enrollment management scores from ACT data. (IR

Applications No. 16). Tallahassee, FL: The Association for Institutional Research.

Sawyer, R.L. (1996). Decision theory models for validating course placement tests. Journal of

Educational Measurement, 33, 271-290.

Sawyer, R. (2010). Usefulness of high school average and ACT scores in making college

admission decisions. (ACT Research Report No. 2010-2). Iowa City, IA: ACT, Inc.

Schmitt, N., Keeney, J., Oswald, F.L., Pleskac, T.J., Billington, A.Q., Sinha, R., & Zorzie, M.

(2009). Prediction of 4-year college student performance using cognitive and

noncognitive predictors and the impact on demographic status of admitted students.

Journal of Applied Psychology, 94(6), 1479-1497.

Soares, J.A. (Ed.) (2012). SAT wars: The case for test-optional college admissions. New York,

NY: Teachers College Press.

Tym, C., McMillion, R., Barone, S., & Webster, J. (2004). First-generation college students: A

literature review. Retrieved October 18, 2012 from

http://www.tgslc.org/pdf/first_generation.pdf.

US Census Bureau. (2003). Median household income: 2003. Retrieved October 23, 2012 from

http://www.census.gov/did/www/saipe/data/statecounty/maps/iy2003/med_hh_inc2003.

pdf.

Woodruff, D.J., & Ziomek, R.L. (2004). High school grade inflation from 1991 to 2003. (ACT

Research Report No. 2004-4). Iowa City, IA: ACT, Inc.

Young, J.W. (2001). Differential validity, differential prediction, and college admission testing:

A comprehensive review and analysis. (College Board Research Report No. 2001-6).

New York: The College Board.

Zwick, R., & Sklar, J.C. (2005). Predicting college grades and degree completion using high

school grades and SAT scores: The role of student ethnicity and first language. American

Educational Research Journal, 42(3), 439-464.

Appendix A

Tables A-1 to A-7

Table A-1

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Four-Year Institutions by Applicant/Enrollment Status and Race/Ethnicity

Enrollment

status

Predictor

variable

Race/

ethnicity

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

White 4,278 126/33,446 21.2 19.0/23.6

Minority 969 19/7,441 17.8 15.9/20.1

HSGPA

White 4,278 126/33,446 3.25 3.00/3.56

Minority 969 19/7,441 2.96 2.70/3.34

Enrolled

students

ACTC

White 901 16/8,064 22.0 16.6/26.2

Minority 187 9/1,413 19.0 15.8/22.7

HSGPA

White 901 16/8,064 3.35 2.97/3.75

Minority 187 9/1,413 3.09 2.74/3.68

Progress

year 1

White 1,086 38/8,064 72 40/90

Minority 206 9/1,413 55 12/85

Progress

year 2

White 1,082 38/8,064 56 25/84

Minority 206 9/1,413 40 10/78

Progress

year 3

White 1,058 37/8,064 48 21/80

Minority 206 9/1,413 34 8/76

Progress

year 4

White 1,054 38/8,064 46 21/80

Minority 206 9/1,413 31 8/74

Bachelor’s

degree

White 901 16/8,064 44 17/80

Minority 187 9/1,413 30 0/70

Year 6

cum GPA

White 463 7/5,258 3.17 2.83/3.53

Minority 78 1/847 2.87 2.28/3.28

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Underrepresented minority students include African American, American Indian, and Hispanic

students. Because some institutions provided data for some but not all of the college outcomes, the descriptive

statistics in the table are based on 61 four-year institutions for ACTC score, HSGPA, and bachelor’s degree

completion by year 6, 50 institutions for the progress to degree outcomes, and 57 institutions for year 6 cumulative

GPA.

Student’s cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6.

Table A-2

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Two-Year Institutions by Applicant/Enrollment Status and Race/Ethnicity

Enrollment

status

Predictor

variable

Race/

ethnicity

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

White 1,648 87/11,747 18.7 17.7/20.7

Minority 458 12/3,396 16.9 15.2/19.1

HSGPA

White 1,648 87/11,747 3.07 2.87/3.25

Minority 458 12/3,396 2.85 2.62/3.12

Enrolled

students

ACTC

White 788 68/6,901 18.8 17.6/20.9

Minority 225 9/1,842 16.8 15.2/19.2

HSGPA

White 788 68/6,901 3.05 2.85/3.31

Minority 225 9/1,842 2.87 2.61/3.24

Progress

year 1

White 740 55/6,413 54 19/78

Minority 210 7/1,647 38 10/75

Progress

year 2

White 740 68/6,448 42 9/63

Minority 210 9/1,667 30 10/56

Progress

year 3

White 739 68/6,387 36 5/57

Minority 209 9/1,674 24 0/49

Associate’s

degree

White 788 68/6,901 15 5/37

Minority 225 9/1,842 8 0/29

Associate’s

degree plus

transfer

White 913 113/6,901 24 7/44

Minority 257 12/1,842 15 4/37

Year 3 cum

GPA

White 317 17/3,466 2.86 2.64/3.16

Minority 60 1/842 2.58 1.53/3.28

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Underrepresented minority students include African American, American Indian, and Hispanic

students. Because some institutions provided data for some but not all of the college outcomes, the descriptive

statistics in the table are based on 43 two-year institutions for ACTC score, HSGPA, and associate’s degree

completion by year 3, 42 institutions for the progress to degree outcomes and year 3 cumulative GPA, and 40

institutions for associate’s degree or transfer to a four-year institution by year 3.

Student’s cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3.

Table A-3

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Four-Year Institutions by Applicant/Enrollment Status and Family Income

Enrollment

status

Predictor

variable

Family

income

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

Low 1,786 53/9,573 19.2 15.9/21.7

Mid 2,760 74/17,237 20.6 16.6/22.9

High 1,725 32/20,132 21.3 17.5/23.9

HSGPA

Low 1,786 53/9,573 3.11 2.79/3.38

Mid 2,760 74/17,237 3.22 2.83/3.49

High 1,725 32/20,132 3.27 2.84/3.56

Enrolled

students

ACTC

Low 375 18/1,488 20.4 15.6/23.6

Mid 492 23/3,425 21.6 16.4/24.8

High 421 9/5,108 22.1 16.7/26.2

HSGPA

Low 375 18/1,488 3.24 2.78/3.74

Mid 492 23/3,425 3.34 2.81/3.72

High 421 9/5,108 3.34 2.82/3.73

Progress

year 1

Low 402 18/1,447 61 18/94

Mid 650 23/3,425 68 27/88

High 421 9/5,108 72 39/92

Progress

year 2

Low 402 18/1,449 45 15/78

Mid 645 23/3,425 56 18/81

High 421 9/5,108 60 27/89

Progress

year 3

Low 402 17/1,410 38 10/74

Mid 633 23/3,425 49 18/78

High 420 9/5,108 53 25/82

Progress

year 4

Low 402 18/1,400 37 10/73

Mid 634 23/3,425 46 18/78

High 420 8/5,108 50 26/82

Bachelor’s

degree

Low 375 18/1,488 33 10/76

Mid 492 23/3,425 43 17/78

High 421 9/5,108 47 24/82

Table A-3 (cont.)

Enrollment

status

Predictor

variable

Family

income

Number of students Mean

Med Min/Max Med Min/Max

Enrolled

students

Year 6

cum GPA

Low 148 5/907 3.04 2.74/3.47

Mid 272 13/2,161 3.13 2.76/3.49

High 209 6/3,475 3.18 2.92/3.42

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Low is for lower-income students (annual family income < $30,000), Mid is for middle-income

students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000). Because some institutions provided data for some but not all of the college outcomes, the

descriptive statistics in the table are based on 61 four-year institutions for ACTC score, HSGPA, and bachelor’s

degree completion by year 6, 50 institutions for the progress to degree outcomes, and 57 institutions for year 6

cumulative GPA.

Student’s cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6.

Table A-4

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Two-Year Institutions by Applicant/Enrollment Status and Family Income

Enrollment

status

Predictor

variable

Family

income

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

Low 862 22/4,902 17.6 16.0/19.8

Mid 937 51/7,061 18.5 17.2/20.4

High 355 47/4,509 19.1 17.7/20.6

HSGPA

Low 862 22/4,902 2.98 2.74/3.15

Mid 937 51/7,061 3.07 2.84/3.25

High 355 47/4,509 3.09 2.86/3.31

Enrolled

students

ACTC

Low 403 18/2,654 17.7 16.1/20.5

Mid 353 39/4,154 18.4 17.1/20.6

High 138 26/2,743 18.9 17.4/20.9

HSGPA

Low 403 18/2,654 2.97 2.72/3.19

Mid 353 39/4,154 3.06 2.82/3.30

High 138 26/2,743 3.07 2.85/3.32

Progress

year 1

Low 383 15/2,399 45 22/74

Mid 351 31/3,841 53 16/79

High 145 23/2,564 59 14/82

Progress

year 2

Low 383 18/2,430 34 9/56

Mid 351 39/3,869 42 9/63

High 145 23/2,567 47 6/67

Progress

year 3

Low 383 18/2,410 28 5/49

Mid 351 39/3,860 36 5/57

High 142 23/2,538 41 0/62

Associate’s

degree

Low 403 18/2,654 12 4/30

Mid 353 39/4,154 17 4/37

High 138 26/2,743 15 2/39

Associate’s

degree plus

transfer

Low 426 46/2,654 17 6/36

Mid 456 85/4,154 26 7/44

High 178 26/2,743 30 9/48

Table A-4 (cont.)

Enrollment

status

Predictor

variable

Family

income

Number of students Mean

Med Min/Max Med Min/Max

Enrolled

students

Year 3

cum GPA

Low 126 7/1,155 2.79 2.40/3.12

Mid 159 14/2,076 2.84 2.65/3.13

High 72 4/1,498 2.85 2.47/3.19

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Low is for lower-income students (annual family income < $30,000), Mid is for middle income

students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000). Because some institutions provided data for some but not all of the college outcomes, the

descriptive statistics in the table are based on 43 two-year institutions for ACTC score, HSGPA, and associate’s

degree completion by year 3, 42 institutions for the progress to degree outcomes and year 3 cumulative GPA, and 40

institutions for associate’s degree or transfer to a four-year institution by year 3.

Student’s cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3.

Table A-5

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Four-Year Institutions by Applicant/Enrollment Status and Gender

Enrollment

status

Predictor

variable Gender

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

Female 3,788 91/21,590 20.4 16.6/22.9

Male 2,447 68/19,922 20.3 15.9/23.3

HSGPA

Female 3,788 91/21,590 3.29 2.91/3.56

Male 2,447 68/19,922 3.11 2.67/3.43

Enrolled

students

ACTC

Female 742 26/5,439 21.7 16.6/25.0

Male 574 24/4,299 21.4 15.4/25.8

HSGPA

Female 742 26/5,439 3.40 2.96/3.76

Male 574 24/4,299 3.21 2.63/3.70

Progress

year 1

Female 878 26/5,439 72 29/91

Male 661 24/4,299 64 26/86

Progress

year 2

Female 877 26/5,439 59 23/85

Male 662 24/4,299 50 18/80

Progress

year 3

Female 872 25/5,439 51 19/83

Male 645 24/4,299 42 16/76

Progress

year 4

Female 868 25/5,439 48 18/83

Male 644 24/4,299 40 17/75

Bachelor’s

degree

Female 742 26/5,439 46 17/81

Male 574 24/4,299 37 14/77

Year 6

cum GPA

Female 394 14/3,484 3.22 2.84/3.68

Male 245 10/2,744 3.00 2.64/3.18

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Because some institutions provided data for some but not all of the college outcomes, the descriptive

statistics in the table are based on 61 four-year institutions for ACTC score, HSGPA, and bachelor’s degree

completion by year 6, 50 institutions for the progress to degree outcomes, and 57 institutions for year 6 cumulative

GPA.

Student’s cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6.

Table A-6

Distributions of Mean ACTC Scores, HSGPAs, College Success Rates, and College GPAs across

Two-Year Institutions by Applicant/Enrollment Status and Gender

Enrollment

status

Predictor

variable Gender

Number of students Mean

Med Min/Max Med Min/Max

Applicant

pool

ACTC

Female 1,274 72/9,661 18.4 16.7/20.3

Male 837 44/6,679 18.3 16.7/20.0

HSGPA

Female 1,274 72/9,661 3.12 2.86/3.29

Male 837 44/6,679 2.93 2.69/3.09

Enrolled

students

ACTC

Female 519 62/5,550 18.4 16.8/20.7

Male 406 29/3,882 18.2 16.8/20.3

HSGPA

Female 519 62/5,550 3.11 2.83/3.33

Male 406 29/3,882 2.92 2.70/3.13

Progress

year 1

Female 512 47/5,167 52 20/76

Male 367 28/3,531 48 15/78

Progress

year 2

Female 512 62/5,184 40 11/60

Male 368 29/3,578 38 5/63

Progress

year 3

Female 511 62/5,134 35 5/54

Male 366 29/3,568 30 2/57

Associate’s

degree

Female 519 62/5,550 15 5/33

Male 406 29/3,882 12 3/40

Associate’s

degree or

transfer

Female 619 72/5,550 23 11/42

Male 473 85/3,882 23 4/43

Year 3

cum GPA

Female 201 18/2,776 2.89 2.55/3.29

Male 146 6/1,898 2.73 2.39/3.46

Note. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade

point average. Because some institutions provided data for some but not all of the college outcomes, the descriptive

statistics in the table are based on 43 two-year institutions for ACTC score, HSGPA, and associate’s degree

completion by year 3, 42 institutions for the progress to degree outcomes and year 3 cumulative GPA, and 40

institutions for associate’s degree or transfer to a four-year institution by year 3.

Student’s cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3.

Table A-7

Comparison of Mean ACTC Scores between Study Sample and ACT-Tested National Sample of

Enrolled Students by Student Demographic Group and Institution Type

Student

demographic

group

Two-year institutions Four-year institutions

Study

sample

National

sample

Study

sample

National

sample

Race/ethnicity

White 18.8 19.4 22.0 23.3

Minority 16.8 16.2 19.0 19.1

Family income range

Low 17.7 17.6 20.4 20.2

Mid 18.4 19.1 21.6 22.2

High 18.9 19.6 22.1 23.6

Gender

Female 18.4 18.6 21.7 22.4

Male 18.2 18.9 21.4 22.9

Note. Results for the study sample are the typical mean ACTC scores across 61 four-year and 43 two-year

institutions (see median values from Tables A-1 to A-6). Results for the national sample are average scores for 2003

ACT-tested high school graduates who enrolled in college in fall 2003 (includes over 581,000 and 191,000 students

who initially enrolled in a four- and two-year institution, respectively), using enrollment data from the National

Student Clearinghouse. Underrepresented minority students include African American, American Indian, and

Hispanic students. Low is for lower-income students (annual family income < $30,000), Mid is for middle-income

students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000).

Appendix B

Figures B-1 to B-18

Figure B-1. Estimated probabilities of degree completion by ACTC score and race/ethnicity.

ACTC = ACT Composite. Bachelor’s degree completion by year 6 at four-year institutions and

associate’s degree completion by year 3 at two-year institutions.

Figure B-2. Estimated probabilities of degree completion by HSGPA and race/ethnicity. HSGPA

= high school grade point average. Bachelor’s degree completion by year 6 at four-year

institutions and associate’s degree completion by year 3 at two-year institutions.

Figure B-3. Estimated probabilities of achieving levels of year 6 cumulative GPA by ACTC

score and race/ethnicity for four-year institutions. ACTC = ACT Composite.

Figure B-4. Estimated probabilities of achieving levels of year 6 cumulative GPA by HSGPA

and race/ethnicity for four-year institutions. HSGPA = high school grade point average.

Figure B-5. Estimated probabilities of achieving levels of year 3 cumulative GPA by ACTC

score and race/ethnicity for two-year institutions. ACTC = ACT Composite.

Figure B-6. Estimated probabilities of achieving levels of year 3 cumulative GPA by HSGPA

and race/ethnicity for two-year institutions. HSGPA = high school grade point average.

Figure B-7. Estimated probabilities of degree completion by ACTC score and family income.

ACTC = ACT Composite. Bachelor’s degree completion by year 6 at four-year institutions and

associate’s degree completion by year 3 at two-year institutions.

Figure B-8. Estimated probabilities of degree completion by HSGPA and family income.

HSGPA = high school grade point average. Bachelor’s degree completion by year 6 at four-year

institutions and associate’s degree completion by year 3 at two-year institutions.

Figure B-9. Estimated probabilities of achieving levels of year 6 cumulative GPA by ACTC

score and family income for four-year institutions. ACTC = ACT Composite.

Figure B-10. Estimated probabilities of achieving levels of year 6 cumulative GPA by HSGPA

and family income for four-year institutions. HSGPA = high school grade point average.

Figure B-11. Estimated probabilities of achieving levels of year 3 cumulative GPA by ACTC

score and family income for two-year institutions. ACTC = ACT Composite.

Figure B-12. Estimated probabilities of achieving levels of year 3 cumulative GPA by HSGPA

and family income for two-year institutions. HSGPA = high school grade point average.

Figure B-13. Estimated probabilities of degree completion by ACTC score and gender. ACTC =

ACT Composite. Bachelor’s degree completion by year 6 at four-year institutions and associate’s

degree completion by year 3 at two-year institutions.

Figure B-14. Estimated probabilities of degree completion by HSGPA and gender. HSGPA =

high school grade point average. Bachelor’s degree completion by year 6 at four-year institutions

and associate’s degree completion by year 3 at two-year institutions.

Figure B-15. Estimated probabilities of achieving levels of year 6 cumulative GPA by ACTC

score and gender for four-year institutions. ACTC = ACT Composite.

Figure B-16. Estimated probabilities of achieving levels of year 6 cumulative GPA by HSGPA

and gender for four-year institutions. HSGPA = high school grade point average.

Figure B-17. Estimated probabilities of achieving levels of year 3 cumulative GPA by ACTC

score and gender for two-year institutions. ACTC = ACT Composite.

Figure B-18. Estimated probabilities of achieving levels of year 3 cumulative GPA by HSGPA

and gender for two-year institutions. HSGPA = high school grade point average.

Appendix C

Tables C-1 to C-6

Table C-1

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Four-

Year Institutions

Outcome variable

Race/

ethnicity

ACTC score HSGPA

< 19 19–25 > 25 <3.00 3.00–3.80 > 3.80

Bachelor’s degree

White -0.02 -0.01 to 0.00 0.00 to 0.01 -0.02 to -0.01 -0.02 to -0.01 -0.01

Minority 0.06 0.03 to 0.06 -0.01 to 0.03 0.02 to 0.05 0.06 to 0.08 0.08

Progress

year 1

White -0.03 to -0.02 -0.02 to 0.00 0.00 -0.03 to -0.02 -0.02 -0.01

Minority 0.06 to 0.08 0.01 to 0.05 -0.01 to 0.00 0.03 to 0.08 0.08 to 0.09 0.07 to 0.08

Progress

year 2

White -0.02 -0.01 to 0.00 0.00 to 0.01 -0.02 to -0.01 -0.02 -0.01

Minority 0.05 to 0.06 0.01 to 0.05 -0.01 to 0.01 0.02 to 0.06 0.07 to 0.08 0.08

Progress

year 3

White -0.02 to -0.01 -0.01 to 0.00 0.00 to 0.01 -0.01 -0.02 to -0.01 -0.01

Minority 0.05 to 0.06 0.02 to 0.05 -0.02 to 0.01 0.02 to 0.06 0.06 to 0.09 0.09

Progress

year 4

White -0.02 -0.01 to 0.00 0.00 to 0.01 -0.01 -0.02 to -0.01 -0.01

Minority 0.05 to 0.06 0.03 to 0.06 -0.02 to 0.02 0.02 to 0.06 0.06 to 0.09 0.09

Year 6 cum GPA

3.00 or higher

White -0.04 to -0.03 -0.03 to 0.00 0.00 to 0.01 -0.04 to -0.02 -0.04 to -0.03 -0.02

Minority 0.05 to 0.11 0.07 to 0.11 0.00 to 0.06 0.03 to 0.13 0.14 to 0.16 0.13 to 0.14

Year 6 cum GPA

3.50 or higher

White -0.01 to 0.00 -0.02 to -0.01 -0.01 to 0.01 -0.01 to 0.00 -0.03 to -0.01 -0.03

Minority 0.01 to 0.04 0.05 to 0.08 0.00 to 0.08 0.00 to 0.03 0.03 to 0.16 0.17 to 0.18

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed

effects parameter estimates from the hierarchical logistic models. The difference between the two models was whether a race/ethnicity indicator was included in

the model (included in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and

75th percentiles for the total group of enrolled students at four-year study institutions. There were 61, 50, and 57 institutions with available outcomes data for

bachelor’s degree completion by year 6, the progress to degree outcomes, and year 6 cumulative GPA, respectively. Students’ cumulative GPAs at degree

completion were included in year 6 GPA analyses for students who graduated with a bachelor’s degree before the end of year 6. Underrepresented minority

students include African American, American Indian, and Hispanic students.

Table C-2

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Two-

Year Institutions

Outcome variable

Race/

ethnicity

ACTC score HSGPA

< 16 16–21 > 21 <2.60 2.60–3.50 > 3.50

Associate’s degree

White -0.01 -0.01 to 0.00 0.00 to 0.02 -0.01 to 0.00 -0.01 -0.01

Minority 0.01 to 0.02 0.01 to 0.02 -0.06 to 0.01 0.01 0.01 to 0.04 0.04 to 0.05

Associate’s degree

or transfer

White -0.01 -0.01 -0.01 to 0.01 -0.01 -0.01 -0.01

Minority 0.02 0.02 -0.03 to 0.01 0.01 to 0.02 0.02 to 0.06 0.06 to 0.08

Progress

year 1

White -0.03 -0.03 to -0.01 0.00 to 0.01 -0.03 -0.03 -0.02

Minority 0.05 to 0.06 0.01 to 0.06 -0.02 to 0.00 0.05 to 0.07 0.08 to 0.10 0.09 to 0.10

Progress

year 2

White -0.02 -0.02 to -0.01 -0.01 to 0.01 -0.02 -0.02 -0.02

Minority 0.04 to 0.05 0.01 to 0.04 -0.03 to 0.01 0.03 to 0.04 0.05 to 0.08 0.08

Progress

year 3

White -0.02 -0.02 to -0.01 -0.01 to 0.01 -0.02 to -0.01 -0.02 -0.02 to -0.01

Minority 0.04 0.01 to 0.04 -0.04 to 0.01 0.02 to 0.03 0.04 to 0.07 0.07

Year 3 cum GPA

3.00 or higher

White -0.02 -0.02 to -0.01 0.00 to 0.01 -0.02 to -0.01 -0.02 -0.01

Minority 0.04 to 0.05 0.03 to 0.05 -0.01 to 0.02 0.03 to 0.05 0.06 to 0.08 0.07 to 0.08

Year 3 cum GPA

3.50 or higher

White -0.01 to 0.00 -0.01 -0.01 to 0.01 -0.01 to 0.00 -0.01 -0.01

Minority 0.01 to 0.02 0.02 to 0.04 0.00 to 0.04 0.01 to 0.02 0.02 to 0.07 0.07 to 0.09

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed

effects parameter estimates from the hierarchical logistic models. The difference between the two models was whether a race/ethnicity indicator was included in

the model (included in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and

75th percentiles for the total group of enrolled students at two-year study institutions. There were 43, 42, 42, and 40 institutions with available outcomes data for

associate’s degree completion by year 3, the progress to degree outcomes, year 3 cumulative GPA, and associate’s degree or transfer to a four-year institution by

year 3, respectively. Students’ cumulative GPAs at degree completion were included in year 3 GPA analyses for students who graduated with an associate’s

degree before the end of year 3. Underrepresented minority students include African American, American Indian, and Hispanic students.

Table C-3

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Four-

Year Institutions

Outcome variable

Family

income

ACTC score HSGPA

< 19 19–25 > 25 <3.00 3.00–3.80 > 3.80

Bachelor’s degree

Low 0.03 to 0.05 0.06 to 0.07 0.06 to 0.07 0.02 to 0.06 0.06 to 0.08 0.08

High -0.05 to -0.04 -0.05 to -0.04 -0.04 to -0.02 -0.05 to -0.03 -0.05 -0.05

Progress

year 1

Low 0.02 to 0.05 0.05 to 0.06 0.02 to 0.05 0.02 to 0.07 0.07 to 0.08 0.06 to 0.07

High -0.04 to -0.03 -0.04 to -0.03 -0.02 to 0.00 -0.06 to -0.03 -0.06 to -0.04 -0.04 to -0.03

Progress

year 2

Low 0.02 to 0.05 0.06 to 0.07 0.04 to 0.07 0.02 to 0.06 0.07 to 0.08 0.08

High -0.05 to -0.04 -0.05 to -0.04 -0.03 to -0.01 -0.06 to -0.03 -0.06 to -0.05 -0.05 to -0.04

Progress

year 3

Low 0.02 to 0.05 0.06 to 0.07 0.05 to 0.07 0.02 to 0.06 0.06 to 0.08 0.08

High -0.04 to -0.03 -0.04 -0.04 to -0.01 -0.05 to -0.02 -0.06 to -0.05 -0.05

Progress

year 4

Low 0.02 to 0.05 0.06 to 0.07 0.06 to 0.07 0.02 to 0.06 0.06 to 0.08 0.08

High -0.04 to -0.03 -0.04 -0.04 to -0.02 -0.05 to -0.02 -0.06 to -0.05 -0.05

Year 6 cum GPA

3.00 or higher

Low 0.01 to 0.02 0.01 to 0.02 0.00 to 0.01 0.01 to 0.04 0.05 to 0.06 0.04 to 0.05

High -0.02 to -0.01 -0.02 to -0.01 -0.01 to 0.00 -0.03 to -0.01 -0.04 to -0.03 -0.03

Year 6 cum GPA

3.50 or higher

Low 0.00 to 0.01 0.02 to 0.03 0.00 to 0.02 0.00 to 0.01 0.02 to 0.07 0.07 to 0.08

High 0.00 -0.01 to 0.00 -0.01 to 0.00 0.00 -0.03 to 0.01 -0.04 to -0.03

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed effects

parameter estimates from the hierarchical logistic models. The difference between the two models was whether an income indicator was included in the model (included

in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and 75th percentiles for the total

group of enrolled students at four-year study institutions. There were 61, 50, and 57 institutions with available outcomes data for bachelor’s degree completion by year 6,

the progress to degree outcomes, and year 6 cumulative GPA, respectively. Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for

students who graduated with a bachelor’s degree before the end of year 6. Low is for lower-income students (annual family income < $30,000) and High is for higher-

income students (annual family income > $60,000). Differences for middle-income students were generally near 0, and therefore are not included in this table.

Table C-4

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Two-

Year Institutions

Outcome variable

Family

income

ACTC score HSGPA

< 16 16–21 > 21 <2.60 2.60–3.50 > 3.50

Associate’s degree

Low 0.01 0.01 to 0.02 -0.01 to 0.01 0.00 to 0.01 0.01 to 0.03 0.03 to 0.04

High -0.01 -0.01 0.00 to 0.06 -0.01 to 0.00 -0.02 to -0.01 -0.03 to -0.02

Associate’s degree

or transfer

Low 0.01 to 0.02 0.03 to 0.04 0.04 to 0.05 0.01 to 0.02 0.02 to 0.05 0.06 to 0.08

High -0.04 to -0.03 -0.05 to -0.04 -0.04 to 0.00 -0.03 to -0.02 -0.07 to -0.03 -0.08 to -0.07

Progress

year 1

Low 0.03 to 0.04 0.04 to 0.05 0.00 to 0.04 0.03 to 0.04 0.05 to 0.06 0.06

High -0.07 to -0.06 -0.07 to -0.04 -0.03 to 0.00 -0.07 to -0.05 -0.08 to -0.07 -0.07 to -0.05

Progress

year 2

Low 0.02 to 0.04 0.04 to 0.05 0.02 to 0.05 0.02 to 0.03 0.04 to 0.06 0.06 to 0.07

High -0.06 -0.06 to -0.05 -0.04 to 0.00 -0.05 to -0.03 -0.08 to -0.05 -0.08 to -0.07

Progress

year 3

Low 0.02 to 0.03 0.03 to 0.04 0.02 to 0.04 0.01 to 0.03 0.03 to 0.06 0.06

High -0.06 to -0.05 -0.06 to -0.04 -0.04 to 0.01 -0.05 to -0.03 -0.07 to -0.05 -0.07 to -0.06

Year 3 cum GPA

3.00 or higher

Low 0.00 0.00 0.00 -0.01 to 0.00 0.00 to 0.02 0.02 to 0.03

High 0.00 to 0.01 0.01 to 0.03 0.01 to 0.03 0.01 0.00 to 0.01 -0.01 to 0.00

Year 3 cum GPA

3.50 or higher

Low 0.00 0.00 0.00 -0.01 -0.01 to 0.02 0.02 to 0.05

High 0.00 to 0.01 0.01 0.00 to 0.02 0.01 -0.01 to 0.01 -0.04 to -0.01

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed effects

parameter estimates from the hierarchical logistic models. The difference between the two models was whether an income indicator was included in the model (included

in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and 75th percentiles for the total

group of enrolled students at two-year study institutions. There were 43, 42, 42, and 40 institutions with available outcomes data for associate’s degree completion by

year 3, the progress to degree outcomes, year 3 cumulative GPA, and associate’s degree or transfer to a four-year institution by year 3, respectively. Students’ cumulative

GPAs at degree completion were included in year 3 GPA analyses for students who graduated with an associate’s degree before the end of year 3. Low is for lower-

income students (annual family income < $30,000) and High is for higher-income students (annual family income > $60,000). Differences for middle-income students

were generally near 0, and therefore are not included in this table.

Table C-5

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Four-

Year Institutions

Outcome variable Gender

ACTC score HSGPA

< 19 19–25 > 25 <3.00 3.00–3.80 > 3.80

Bachelor’s degree

Female -0.03 to -0.01 -0.04 to -0.03 -0.05 to -0.04 -0.01 to 0.00 -0.02 to -0.01 -0.02

Male 0.02 to 0.04 0.04 to 0.06 0.05 to 0.06 0.00 to 0.01 0.01 to 0.03 0.03

Progress

year 1

Female -0.04 to 0.00 -0.04 -0.04 to -0.01 -0.02 to 0.00 -0.02 to -0.01 -0.01

Male 0.02 to 0.06 0.05 to 0.06 0.02 to 0.05 0.00 to 0.02 0.02 to 0.03 0.02

Progress

year 2

Female -0.04 to -0.01 -0.05 to -0.04 -0.04 to -0.02 -0.02 to 0.00 -0.02 to -0.01 -0.01

Male 0.02 to 0.05 0.05 to 0.06 0.03 to 0.06 0.00 to 0.02 0.02 0.02

Progress

year 3

Female -0.03 to -0.01 -0.05 to -0.04 -0.05 to -0.03 -0.01 -0.02 to -0.01 -0.02 to -0.01

Male 0.02 to 0.05 0.05 to 0.06 0.04 to 0.06 0.00 to 0.01 0.02 to 0.03 0.02 to 0.03

Progress

year 4

Female -0.03 to -0.01 -0.05 to -0.03 -0.05 to -0.03 -0.01 -0.01 -0.01

Male 0.02 to 0.04 0.05 to 0.06 0.04 to 0.06 0.00 to 0.01 0.02 0.02

Year 6 cum GPA

3.00 or higher

Female -0.07 to 0.00 -0.09 to -0.08 -0.07 to -0.02 -0.06 to -0.01 -0.06 to -0.04 -0.04 to -0.03

Male 0.04 to 0.11 0.11 to 0.13 0.03 to 0.10 0.01 to 0.06 0.07 to 0.08 0.07

Year 6 cum GPA

3.50 or higher

Female -0.02 to 0.00 -0.09 to -0.03 -0.10 to -0.04 -0.01 to 0.00 -0.04 to -0.02 -0.04

Male 0.01 to 0.04 0.05 to 0.13 0.06 to 0.13 0.00 to 0.01 0.02 to 0.08 0.09

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed

effects parameter estimates from the hierarchical logistic models. The difference between the two models was whether a gender indicator was included in the

model (included in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and 75th

percentiles for the total group of enrolled students at four-year study institutions. There were 61, 50, and 57 institutions with available outcomes data for

bachelor’s degree completion by year 6, the progress to degree outcomes, and year 6 cumulative GPA, respectively. Students’ cumulative GPAs at degree

completion were included in year 6 GPA analyses for students who graduated with a bachelor’s degree before the end of year 6.

Table C-6

Differences in Probabilities of Success between Total-Group and Group-Specific Models based on ACTC Score or HSGPA for Two-

Year Institutions

Outcome variable Gender

ACTC score HSGPA

< 16 16–21 > 21 <2.60 2.60–3.50 > 3.50

Associate’s degree

Female 0.00 -0.02 to 0.00 -0.07 to -0.02 0.00 0.00 -0.02 to -0.01

Male 0.00 to 0.01 0.01 to 0.03 0.03 to 0.10 0.00 0.00 to 0.01 0.02 to 0.03

Associate’s degree

or transfer

Female 0.01 -0.01 to 0.00 -0.05 to -0.01 0.01 to 0.02 0.01 to 0.02 -0.01 to 0.01

Male 0.00 to 0.01 0.00 to 0.01 0.02 to 0.06 -0.02 to -0.01 -0.02 to -0.01 -0.01 to 0.00

Progress

year 1

Female 0.01 to 0.02 -0.03 to 0.00 -0.03 to -0.01 0.03 0.00 to 0.03 -0.02 to 0.00

Male -0.02 to 0.00 0.00 to 0.03 0.02 to 0.04 -0.03 -0.02 to 0.00 0.00 to 0.01

Progress

year 2

Female 0.00 to 0.01 -0.03 to 0.00 -0.04 to -0.03 0.02 0.00 to 0.02 -0.02 to -0.01

Male -0.01 to 0.01 0.01 to 0.04 0.04 to 0.05 -0.02 -0.01 to 0.01 0.01 to 0.02

Progress

year 3

Female 0.00 to 0.01 -0.02 to 0.00 -0.04 to -0.03 0.01 0.00 to 0.01 -0.01 to 0.00

Male 0.00 to 0.01 0.01 to 0.04 0.04 to 0.06 -0.01 -0.01 to 0.01 0.01 to 0.02

Year 3 cum GPA

3.00 or higher

Female -0.02 to 0.00 -0.06 to -0.03 -0.06 to -0.02 -0.01 -0.03 to -0.02 -0.03 to -0.02

Male 0.01 to 0.04 0.05 to 0.09 0.05 to 0.10 0.01 to 0.02 0.02 to 0.05 0.05 to 0.06

Year 3 cum GPA

3.50 or higher

Female 0.00 -0.04 to -0.01 -0.08 to -0.04 0.00 -0.01 to 0.00 -0.02 to -0.01

Male 0.00 to 0.01 0.02 to 0.06 0.07 to 0.12 0.00 0.00 to 0.03 0.03 to 0.06

Note. Negative differences indicate underprediction and positive differences indicate overprediction. The probabilities of success were estimated using the fixed

effects parameter estimates from the hierarchical logistic models. The difference between the two models was whether a gender indicator was included in the

model (included in the group-specific model, but not in the total-group model). The cutoffs used in the table for ACTC score and HSGPA are the 25th and 75th

percentiles for the total group of enrolled students at two-year study institutions. There were 43, 42, 42, and 40 institutions with available outcomes data for

associate’s degree completion by year 3, the progress to degree outcomes, year 3 cumulative GPA, and associate’s degree or transfer to a four-year institution by

year 3, respectively. Students’ cumulative GPAs at degree completion were included in year 3 GPA analyses for students who graduated with an associate’s

degree before the end of year 3.

Appendix D

Tables D-1 to D-12

Table D-1

Results for Bachelor’s Degree Completion and Progress to Degree at Four-Year Institutions based on ACTC Score and HSGPA

Models by Race/Ethnicity at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

ACTC White 58 0.51 0.44/0.62 63 53/82 20 -1/65 56 50/77 77 0/100

(25) Minority 58 0.48 0.31/0.58 73 48/90 46 0/80 52 36/64 95 0/100

HSGPA White 56 0.51 0.48/0.58 65 58/77 21 0/51 59 52/77 65 2/95

(3.57) Minority 56 0.43 0.26/0.50 69 55/87 40 2/75 49 29/64 81 5/99

ACTC &

HSGPA

White 61 0.51 0.44/0.55 65 58/83 24 0/66 59 51/77 67 2/99

Minority 61 0.45 0.33/0.54 73 57/89 46 3/79 54 39/63 85 9/100

Progress to degree year 1

ACTC White 44 0.54 0.50/0.62 70 62/87 2 0/33 72 62/87 17 0/75

(18) Minority 44 0.45 0.36/0.55 68 55/87 16 0/73 62 48/78 46 0/95

HSGPA White 48 0.52 0.46/0.68 73 64/88 5 0/34 75 65/88 23 0/82

(2.80) Minority 48 0.43 0.26/0.53 68 61/84 13 0/71 60 33/80 39 0/89

ACTC & White 48 0.51 0.47/0.55 74 67/85 5 0/37 75 67/86 22 0/68

HSGPA

Minority 48 0.45 0.36/0.63 71 62/88 18 0/74 65 49/79 48 0/92

Progress to degree year 2

ACTC White 49 0.52 0.50/0.56 65 58/80 9 0/46 64 59/80 44 0/88

(20) Minority 49 0.47 0.32/0.56 67 52/88 32 0/77 58 41/71 74 0/99

HSGPA White 50 0.51 0.48/0.62 69 62/80 13 0/45 67 57/80 42 1/93

(3.13) Minority 50 0.43 0.26/0.49 67 57/87 29 1/76 56 33/72 65 2/96

ACTC & White 50 0.51 0.49/0.53 70 62/78 14 1/47 68 61/79 44 3/82

HSGPA

Minority 50 0.46 0.35/0.51 69 57/89 34 3/77 59 46/70 69 14/97

Table D-1 (cont.)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

ACTC White 50 0.52 0.48/0.56 63 56/78 15 0/55 60 55/78 63 0/95

(22) Minority 50 0.46 0.31/0.57 70 51/90 39 0/81 55 38/68 89 0/100

HSGPA White 50 0.51 0.49/0.59 67 60/77 19 0/54 63 54/77 53 1/94

(3.39) Minority 50 0.43 0.25/0.48 69 55/90 38 1/80 52 30/70 76 4/99

ACTC & White 50 0.51 0.48/0.54 68 60/79 20 1/56 64 58/76 55 4/90

HSGPA

Minority 50 0.45 0.33/0.51 72 56/91 41 4/81 56 43/68 80 16/98

Progress to degree year 4

ACTC White 48 0.52 0.48/0.57 64 55/79 19 -1/57 59 54/74 68 0/97

(24) Minority 48 0.47 0.31/0.60 73 47/90 45 0/80 55 37/67 92 0/100

HSGPA White 49 0.51 0.47/0.57 66 59/78 22 0/56 62 53/77 56 3/94

(3.46) Minority 49 0.43 0.24/0.50 71 54/89 42 1/80 51 27/68 78 6/97

ACTC & White 50 0.51 0.47/0.55 67 59/79 22 1/58 62 55/76 60 5/92

HSGPA

Minority 50 0.45 0.31/0.54 73 55/90 46 4/80 55 40/67 85 16/99

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 61 and 50 institutions with available outcomes data for bachelor’s degree completion and

progress to degree analyses, respectively. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point

average. Underrepresented minority students include African American, American Indian, and Hispanic students.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-2

Results for Associate’s Degree Completion and Progress to Degree at Two-Year Institutions based on ACTC Score and HSGPA

Models by Race/Ethnicity at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

ACTC White 25 0.50 0.48/0.52 81 64/92 61 28/84 53 49/60 99 90/100

(29) Minority 25 0.55 0.52/0.60 86 71/96 72 41/92 50 26/65 100 94/100

HSGPA White 5 0.51 0.50/0.51 70 67/73 39 31/45 53 51/55 90 81/94

(3.92) Minority 5 0.49 0.47/0.50 75 72/79 50 42/59 50 49/53 95 85/96

ACTC &

HSGPA

White 25 0.49 0.45/0.51 81 63/92 61 26/83 52 48/57 99 80/100

Minority 25 0.47 0.25/0.55 87 73/97 75 44/93 49 24/58 100 86/100

Associate’s degree completion or transfer to a four-year institution by year 3

ACTC White 38 0.51 0.48/0.53 76 61/87 51 19/73 55 48/60 98 72/100

(27) Minority 38 0.52 0.49/0.55 85 66/92 69 30/85 54 34/67 100 82/100

HSGPA White 14 0.51 0.49/0.53 68 61/74 33 20/47 55 51/61 84 61/96

(3.75) Minority 14 0.45 0.39/0.50 74 65/86 48 29/71 49 39/56 91 70/100

ACTC & White 38 0.50 0.49/0.52 76 61/87 51 21/73 54 50/62 96 62/100

HSGPA

Minority 38 0.47 0.34/0.51 84 66/92 68 31/85 52 33/63 100 73/100

Progress to degree year 1

ACTC White 41 0.54 0.50/0.61 65 57/77 11 0/54 66 54/74 48 0/99

(19) Minority 41 0.48 0.39/0.61 69 58/86 31 0/71 60 51/70 69 0/97

HSGPA White 41 0.52 0.48/0.60 65 58/78 10 0/34 66 60/78 44 1/85

(3.03) Minority 41 0.43 0.28/0.51 65 59/84 27 0/71 53 34/76 56 3/95

ACTC & White 42 0.51 0.43/0.57 67 61/78 12 0/52 68 56/78 45 0/99

HSGPA

Minority 42 0.47 0.36/0.55 70 60/88 31 0/77 61 45/75 66 2/100

Table D-2 (cont.)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

ACTC White 41 0.52 0.49/0.57 63 59/73 20 0/47 60 52/67 72 14/96

(21) Minority 41 0.50 0.44/0.59 71 57/86 41 6/71 58 51/66 92 25/100

HSGPA White 41 0.51 0.50/0.57 65 59/74 22 1/47 61 55/69 67 18/95

(3.38) Minority 41 0.43 0.32/0.52 70 55/85 40 6/69 50 25/66 82 26/100

ACTC & White 41 0.51 0.49/0.56 66 62/74 23 1/48 62 55/69 67 19/94

HSGPA

Minority 41 0.47 0.38/0.55 71 57/85 41 7/70 55 49/68 85 31/99

Progress to degree year 3

ACTC White 41 0.51 0.49/0.56 66 57/77 30 1/54 57 53/63 86 32/100

(23) Minority 41 0.51 0.44/0.60 76 59/88 52 12/75 57 47/64 97 44/100

HSGPA White 36 0.51 0.49/0.54 66 58/73 30 3/45 57 53/66 79 36/98

(3.62) Minority 36 0.43 0.32/0.52 73 57/87 45 14/74 49 32/62 89 44/100

ACTC & White 41 0.51 0.49/0.54 68 59/77 32 3/55 59 53/67 81 37/98

HSGPA

Minority 41 0.48 0.39/0.53 75 59/87 50 16/74 55 48/64 93 48/100

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 43 and 42 institutions with available outcomes data for associate’s degree completion and

progress to degree analyses, respectively. There were 40 institutions with data available for associate’s degree or transfer to a four-year institution by year 3. Med

= Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Underrepresented minority students

include African American, American Indian, and Hispanic students.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-3

Results for Achieving Levels of Year 6 College Cumulative GPA at Four-Year Institutions based on ACTC Score and HSGPA Models

by Race/Ethnicity at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC White 57 0.55 0.52/0.62 66 62/73 7 0/24 70 62/76 36 16/76

(20) Minority 57 0.41 0.35/0.54 70 61/79 37 7/58 54 48/64 65 35/91

HSGPA White 57 0.53 0.50/0.73 71 63/76 13 -4/33 73 62/85 39 18/68

(3.17) Minority 57 0.34 0.27/0.50 70 59/79 37 9/58 51 39/63 59 34/84

ACTC &

HSGPA

White 57 0.53 0.50/0.66 71 67/76 15 -1/35 73 64/82 42 15/71

Minority 57 0.38 0.33/0.50 75 63/83 45 11/66 57 47/65 71 42/88

3.50 or higher

ACTC White 57 0.53 0.50/0.61 77 59/87 52 12/74 61 56/70 91 76/98

(27) Minority 57 0.45 0.39/0.55 91 71/96 82 40/91 52 37/66 98 89/100

HSGPA White 40 0.52 0.50/0.69 78 55/83 51 -1/65 58 53/73 84 61/96

(3.86) Minority 40 0.32 0.23/0.45 90 71/94 81 42/89 36 25/49 95 74/99

ACTC & White 57 0.51 0.49/0.62 80 59/89 58 8/77 63 56/70 88 65/98

HSGPA

Minority 57 0.40 0.31/0.49 93 72/97 85 42/93 50 33/70 97 84/100

Note. All statistics presented in the table are evaluated at the institution-specific optimal total-group selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6. There were 57 institutions with data available for the year 6 college GPA analyses. Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Underrepresented minority students include African

American, American Indian, and Hispanic students.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-4

Results for Achieving Levels of Year 3 College Cumulative GPA at Two-Year Institutions based on ACTC Score and HSGPA Models

by Race/Ethnicity at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Race/

ethnicity

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC White 42 0.53 0.49/0.59 64 60/68 17 2/30 62 56/73 62 27/88

(20) Minority 42 0.48 0.43/0.54 71 61/80 40 17/61 58 51/67 82 48/98

HSGPA White 42 0.52 0.49/0.55 67 63/70 22 3/35 63 59/72 63 31/80

(3.30) Minority 42 0.41 0.33/0.49 72 59/81 42 18/62 54 39/64 74 39/91

ACTC &

HSGPA

White 42 0.51 0.49/0.54 68 63/71 24 4/36 65 60/74 61 31/79

Minority 42 0.46 0.39/0.50 75 62/81 44 19/61 59 48/71 78 46/92

3.50 or higher

ACTC White 42 0.52 0.45/0.56 81 70/87 62 37/74 59 50/65 97 84/100

(26) Minority 42 0.50 0.44/0.56 90 80/94 81 61/88 55 34/76 99 89/100

HSGPA White 5 0.52 0.51/0.53 77 72/80 53 42/60 53 52/56 93 85/96

(3.98) Minority 5 0.39 0.39/0.42 87 81/89 74 65/78 42 39/45 97 89/98

ACTC & White 42 0.50 0.45/0.54 83 72/88 65 40/77 58 49/65 95 84/99

HSGPA

Minority 42 0.44 0.28/0.51 91 81/95 81 60/90 54 27/66 99 86/100

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3. There were 42 institutions with data available for the year 3 college GPA analyses. Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Underrepresented minority students include African

American, American Indian, and Hispanic students.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-5

Results for Bachelor’s Degree Completion and Progress to Degree at Four-Year Institutions based on ACTC Score and HSGPA

Models by Family Income at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

ACTC Low 58 0.44 0.39/0.49 69 52/90 39 0/79 49 42/65 90 0/100

(25) Mid 58 0.50 0.47/0.53 63 55/82 23 0/65 56 49/71 82 0/100

High 58 0.55 0.53/0.57 62 52/77 14 0/55 60 54/77 78 0/99

HSGPA Low 56 0.43 0.38/0.47 68 57/85 35 1/70 51 42/71 73 4/97

(3.57) Mid 56 0.49 0.46/0.53 65 59/77 21 0/55 58 51/76 66 3/96

High 56 0.55 0.53/0.59 64 57/79 14 0/41 63 56/80 62 2/95

ACTC &

HSGPA

Low 61 0.43 0.40/0.49 70 58/89 39 1/78 52 43/71 77 3/100

Mid 61 0.49 0.47/0.52 66 59/83 26 0/65 58 51/75 68 3/99

High 61 0.55 0.53/0.58 65 57/79 18 0/56 64 56/79 63 2/99

Progress to degree year 1

ACTC Low 44 0.47 0.42/0.52 68 57/83 9 0/63 62 52/76 34 0/93

(18) Mid 44 0.53 0.50/0.59 70 62/83 3 0/52 71 62/83 20 0/84

High 44 0.56 0.52/0.64 72 62/88 1 0/40 75 66/88 14 0/76

HSGPA Low 48 0.44 0.36/0.53 70 61/80 10 0/59 63 47/78 31 0/90

(2.80) Mid 48 0.49 0.47/0.55 73 66/84 6 0/47 73 63/84 24 0/88

High 48 0.55 0.51/0.63 74 67/88 3 0/34 78 69/88 21 0/83

ACTC & Low 48 0.45 0.40/0.67 71 62/83 11 0/64 66 56/81 35 0/91

HSGPA

Mid 48 0.50 0.48/0.55 74 67/84 7 0/55 74 66/84 26 0/84

High 48 0.54 0.51/0.59 76 69/87 4 0/41 79 70/87 21 0/73

Table D-5 (cont.)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

ACTC Low 49 0.46 0.40/0.51 64 54/85 21 0/70 55 49/71 57 0/97

(20) Mid 49 0.51 0.49/0.57 64 57/82 9 0/63 63 57/76 50 0/92

High 49 0.56 0.53/0.61 67 59/82 4 0/50 68 62/82 39 0/86

HSGPA Low 50 0.43 0.36/0.49 66 58/83 24 1/67 57 43/71 50 2/96

(3.13) Mid 50 0.49 0.47/0.54 69 60/79 14 0/57 65 57/77 44 2/96

High 50 0.55 0.53/0.60 69 63/82 8 0/46 70 62/82 40 1/92

ACTC & Low 50 0.44 0.40/0.49 67 59/86 25 1/71 58 51/68 54 10/96

HSGPA

Mid 50 0.50 0.48/0.53 69 61/83 17 1/65 66 60/74 47 5/91

High 50 0.55 0.52/0.58 70 63/80 9 0/50 71 65/81 41 3/83

Progress to degree year 3

ACTC Low 50 0.45 0.40/0.51 67 55/89 31 0/79 53 45/68 76 0/99

(22) Mid 50 0.51 0.48/0.55 63 54/82 15 0/63 61 53/75 64 0/96

High 50 0.55 0.52/0.60 64 56/80 9 0/51 65 58/80 60 0/94

HSGPA Low 50 0.43 0.37/0.48 67 57/88 31 1/76 54 40/68 64 4/98

(3.39) Mid 50 0.50 0.47/0.54 66 59/80 20 0/60 61 52/74 55 3/97

High 50 0.55 0.53/0.60 67 60/78 13 0/49 67 60/79 52 1/95

ACTC & Low 50 0.44 0.40/0.47 68 57/88 34 3/77 55 49/66 68 11/97

HSGPA

Mid 50 0.50 0.48/0.52 68 59/83 21 1/65 62 57/73 58 7/93

High 50 0.55 0.53/0.57 68 60/78 14 0/51 68 62/78 53 4/89

Table D-5 (cont.)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 4

ACTC Low 48 0.44 0.39/0.50 68 52/89 35 0/79 51 42/66 79 0/99

(24) Mid 48 0.51 0.48/0.55 63 54/81 22 0/62 58 52/71 70 0/98

High 48 0.55 0.53/0.60 62 55/77 14 0/50 64 57/77 65 0/97

HSGPA Low 49 0.43 0.37/0.46 68 57/85 33 1/72 53 41/66 68 6/97

(3.46) Mid 49 0.50 0.47/0.52 67 58/79 22 1/58 60 52/74 60 4/97

High 49 0.55 0.53/0.60 66 60/78 15 0/47 66 57/79 57 2/95

ACTC & Low 50 0.44 0.40/0.47 69 57/89 35 2/77 54 46/65 73 9/98

HSGPA

Mid 50 0.50 0.47/0.52 67 59/82 24 1/63 60 54/73 62 6/94

High 50 0.55 0.53/0.57 67 60/77 17 0/50 66 60/78 58 4/92

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 61 and 50 institutions with available outcomes data for bachelor’s degree completion and

progress to degree analyses, respectively. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point

average. Low is for lower-income students (annual family income < $30,000), Mid is for middle-income students (annual family income between $30,000 and

$60,000), and High is for higher-income students (annual family income > $60,000).

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-6

Results for Associates’s Degree Completion and Progress to Degree at Two-Year Institutions based on ACTC Score and HSGPA

Models by Family Income at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

ACTC Low 25 0.49 0.47/0.54 84 71/94 68 42/89 52 35/58 100 94/100

(29) Mid 25 0.51 0.50/0.55 80 63/92 60 26/84 56 51/64 99 91/100

High 25 0.47 0.46/0.52 79 62/92 57 23/83 49 32/57 99 88/100

HSGPA Low 5 0.45 0.44/0.45 75 72/78 51 44/57 47 44/48 93 85/96

(3.92) Mid 5 0.53 0.52/0.53 70 67/72 38 30/44 54 53/57 89 82/93

High 5 0.52 0.52/0.53 69 67/72 36 30/43 54 53/57 88 80/94

ACTC &

HSGPA

Low 25 0.46 0.39/0.52 84 71/95 69 42/89 50 38/59 100 86/100

Mid 25 0.51 0.44/0.53 80 63/92 60 26/84 55 50/59 99 81/100

High 25 0.49 0.35/0.52 79 62/91 57 23/82 51 34/58 98 76/100

Associate’s degree completion or transfer to a four-year institution by year 3

ACTC Low 38 0.46 0.43/0.49 81 67/92 63 34/84 50 38/57 99 81/100

(27) Mid 38 0.53 0.49/0.56 75 61/88 50 18/76 57 47/63 98 73/100

High 38 0.54 0.50/0.57 71 58/84 41 12/69 58 44/64 97 69/100

HSGPA Low 14 0.43 0.42/0.44 73 66/79 47 33/58 46 43/52 89 68/96

(3.75) Mid 14 0.52 0.51/0.54 68 62/73 34 21/46 56 53/63 85 62/99

High 14 0.57 0.55/0.59 65 58/69 26 13/37 61 59/66 83 60/95

ACTC & Low 38 0.44 0.37/0.46 81 67/92 62 34/83 48 37/55 98 70/100

HSGPA

Mid 38 0.52 0.47/0.53 76 62/88 51 22/76 56 46/65 96 63/100

High 38 0.55 0.48/0.58 70 59/85 39 14/70 59 50/68 94 60/100

Table D-6 (cont.)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 1

ACTC Low 41 0.47 0.44/0.52 66 57/82 25 0/63 60 50/67 61 0/99

(19) Mid 41 0.54 0.49/0.59 65 57/79 11 0/58 67 44/74 50 0/100

High 41 0.57 0.50/0.63 65 58/78 6 -1/57 70 50/78 42 0/98

HSGPA Low 41 0.45 0.39/0.47 64 59/78 19 0/55 58 45/73 52 2/93

(3.03) Mid 41 0.52 0.50/0.54 65 63/79 12 0/44 66 58/79 48 1/88

High 41 0.57 0.55/0.59 66 59/82 7 -1/32 72 64/82 42 1/85

ACTC & Low 42 0.45 0.33/0.47 68 60/82 25 0/64 61 45/72 57 0/100

HSGPA

Mid 42 0.52 0.37/0.53 68 63/79 15 0/57 69 47/79 48 0/100

High 42 0.56 0.46/0.58 69 60/82 8 0/51 73 54/83 42 0/98

Progress to degree year 2

ACTC Low 41 0.47 0.43/0.49 68 56/83 34 4/66 55 43/60 79 21/100

(21) Mid 41 0.53 0.49/0.57 64 60/76 20 1/51 61 53/68 75 14/96

High 41 0.56 0.51/0.60 62 57/72 13 -2/41 64 56/71 69 11/96

HSGPA Low 41 0.44 0.40/0.46 67 56/82 33 5/63 53 22/62 75 21/100

(3.38) Mid 41 0.52 0.49/0.54 65 61/77 21 1/53 61 51/69 69 18/97

High 41 0.57 0.56/0.60 64 58/72 14 0/42 68 60/74 65 19/93

ACTC & Low 41 0.45 0.38/0.47 69 57/83 34 6/66 56 46/63 74 26/96

HSGPA

Mid 41 0.51 0.48/0.54 67 62/78 22 1/54 63 52/70 68 18/91

High 41 0.56 0.52/0.60 66 59/73 16 0/43 68 58/74 62 17/91

Table D-6 (cont.)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

ACTC Low 41 0.46 0.44/0.50 72 58/86 43 12/71 53 37/58 92 42/100

(23) Mid 41 0.52 0.49/0.56 66 58/80 30 3/61 58 52/63 88 34/99

High 41 0.55 0.50/0.59 62 57/75 20 -1/51 60 52/67 83 30/99

HSGPA Low 36 0.44 0.41/0.46 71 56/82 41 11/64 50 42/60 85 43/99

(3.62) Mid 36 0.52 0.50/0.54 67 59/75 31 5/50 59 52/67 80 36/98

High 36 0.57 0.55/0.59 64 60/71 20 0/41 63 57/72 76 37/98

ACTC & Low 41 0.45 0.41/0.48 73 58/86 44 12/71 53 40/60 85 46/100

HSGPA

Mid 41 0.51 0.47/0.53 68 59/81 32 5/62 60 48/68 80 39/98

High 41 0.56 0.49/0.58 65 61/76 24 1/50 64 55/72 75 33/98

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 43 and 42 institutions with available outcomes data for associate’s degree completion and

progress to degree analyses, respectively. There were 40 institutions with data available for associate’s degree or transfer to a four-year institution by year 3. Med

= Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Low is for lower-income students (annual

family income < $30,000), Mid is for middle-income students (annual family income between $30,000 and $60,000), and High is for higher-income students

(annual family income > $60,000).

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-7

Results for Achieving Levels of Year 6 College Cumulative GPA at Four-Year Institutions based on ACTC Score and HSGPA Models

by Family Income at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC Low 57 0.50 0.42/0.55 67 61/75 19 6/46 65 57/71 53 24/89

(20) Mid 57 0.52 0.50/0.55 66 62/72 10 3/30 67 61/75 40 15/82

High 57 0.54 0.51/0.57 67 63/75 7 2/23 71 63/78 33 14/76

HSGPA Low 57 0.45 0.37/0.51 70 62/75 23 9/46 63 56/71 49 28/80

(3.17) Mid 57 0.49 0.47/0.53 70 64/75 17 4/36 69 59/77 40 18/76

High 57 0.53 0.51/0.59 71 66/76 12 3/31 74 63/80 36 16/77

ACTC &

HSGPA

Low 57 0.48 0.40/0.53 72 64/79 30 9/53 67 57/75 58 33/82

Mid 57 0.49 0.47/0.52 72 65/75 20 5/38 70 62/77 45 15/76

High 57 0.52 0.50/0.56 72 68/77 14 4/32 74 66/80 39 16/70

3.50 or higher

ACTC Low 57 0.51 0.43/0.54 84 71/93 68 41/87 58 51/66 96 84/100

(27) Mid 57 0.52 0.47/0.56 79 66/90 56 25/79 60 54/68 93 72/100

High 57 0.53 0.48/0.60 77 68/87 51 30/72 62 53/71 89 69/98

HSGPA Low 40 0.42 0.37/0.48 83 71/92 67 38/84 48 39/58 89 75/98

(3.86) Mid 40 0.50 0.47/0.52 79 67/86 56 22/72 56 49/64 85 59/97

High 40 0.53 0.51/0.58 77 67/83 50 24/65 59 53/65 84 59/95

ACTC & Low 57 0.47 0.42/0.50 87 70/94 73 37/87 57 49/64 93 79/99

HSGPA

Mid 57 0.50 0.48/0.51 82 66/91 62 20/82 60 55/66 90 61/98

High 57 0.51 0.48/0.54 80 69/88 56 25/75 64 56/70 85 66/98

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6. There were 57 institutions with data available for the year 6 college GPA analyses. Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Low is for lower-income students (annual family

income < $30,000), Mid is for middle-income students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000).

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-8

Results for Achieving Levels of Year 3 College Cumulative GPA at Two-Year Institutions based on ACTC Score and HSGPA Models

by Family Income at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV)

Family

income

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC Low 42 0.52 0.48/0.57 66 61/74 25 6/45 62 58/67 74 36/94

(20) Mid 42 0.53 0.51/0.57 66 62/69 20 4/34 64 59/71 65 33/91

High 42 0.49 0.45/0.55 65 60/68 20 4/33 61 53/71 57 27/84

HSGPA Low 42 0.49 0.47/0.51 68 61/74 29 6/45 60 57/64 69 34/88

(3.30) Mid 42 0.51 0.50/0.53 68 63/71 23 6/38 65 61/70 63 30/82

High 42 0.50 0.47/0.53 68 64/72 24 6/41 63 57/72 59 31/83

ACTC &

HSGPA

Low 42 0.50 0.47/0.52 70 62/76 31 8/49 64 58/70 72 40/87

Mid 42 0.51 0.48/0.52 70 64/73 26 7/39 66 61/72 62 33/81

High 42 0.48 0.45/0.52 70 64/72 26 7/40 64 58/73 57 29/76

3.50 or higher

ACTC Low 42 0.52 0.47/0.57 86 75/90 71 48/79 59 50/65 98 88/100

(26) Mid 42 0.53 0.49/0.56 83 72/88 65 39/76 61 49/66 97 84/100

High 42 0.50 0.48/0.55 82 75/87 63 46/74 58 41/68 95 81/100

HSGPA Low 5 0.45 0.42/0.46 80 78/83 61 57/67 47 42/49 95 90/97

(3.98) Mid 5 0.52 0.50/0.53 75 74/80 49 48/61 53 51/55 92 86/95

High 5 0.53 0.52/0.57 77 74/80 51 45/60 54 53/61 90 84/95

ACTC & Low 42 0.49 0.44/0.54 86 75/91 72 49/82 57 49/64 97 88/100

HSGPA

Mid 42 0.50 0.47/0.53 84 72/89 66 41/78 60 52/64 95 84/99

High 42 0.50 0.47/0.54 83 76/88 64 49/77 60 52/65 94 81/99

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3. There were 42 institutions with data available for the year 3 college GPA analyses.Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average. Low is for lower-income students (annual family

income < $30,000), Mid is for middle-income students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000).

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

Table D-9

Results for Bachelor’s Degree Completion and Progress to Degree at Four-Year Institutions based on ACTC Score and HSGPA

Models by Gender at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

ACTC Female 58 0.55 0.50/0.59 63 53/83 19 0/65 61 55/76 85 0/100

(25) Male 58 0.45 0.41/0.52 66 54/86 31 0/72 51 45/66 82 0/99

HSGPA Female 56 0.51 0.49/0.54 65 57/78 19 0/53 60 52/79 61 2/95

(3.57) Male 56 0.48 0.44/0.51 67 59/81 29 0/62 55 46/74 73 4/98

ACTC &

HSGPA

Female 61 0.52 0.49/0.55 66 57/83 21 0/65 60 55/79 65 2/99

Male 61 0.47 0.43/0.52 67 59/86 31 0/72 55 47/73 76 4/99

Progress to degree year 1

ACTC Female 44 0.55 0.49/0.62 72 62/86 2 0/55 74 65/86 20 0/90

(18) Male 44 0.47 0.42/0.52 68 58/82 6 0/62 65 55/80 26 0/87

HSGPA Female 48 0.51 0.49/0.55 74 66/88 4 0/48 75 58/88 20 0/88

(2.80) Male 48 0.49 0.46/0.52 71 63/82 8 0/54 69 58/82 33 0/90

ACTC & Female 48 0.52 0.49/0.60 75 67/86 4 0/56 77 67/86 21 0/87

HSGPA

Male 48 0.48 0.44/0.62 73 64/83 10 0/62 70 61/81 32 0/87

Progress to degree year 2

ACTC Female 49 0.55 0.50/0.59 66 59/82 7 0/62 67 61/80 52 0/95

(20) Male 49 0.46 0.42/0.50 64 55/86 17 0/70 58 50/73 51 0/94

HSGPA Female 50 0.51 0.49/0.55 69 63/81 11 0/58 67 54/81 38 1/95

(3.13) Male 50 0.48 0.45/0.52 68 58/83 19 0/65 62 54/75 53 2/96

ACTC & Female 50 0.52 0.49/0.55 70 63/83 12 0/64 68 61/80 43 4/93

HSGPA

Male 50 0.48 0.44/0.50 69 59/86 22 1/70 63 58/73 54 8/94

100

Table D-9 (cont.)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

ACTC Female 50 0.55 0.52/0.61 64 57/82 15 0/64 64 59/80 66 0/97

(22) Male 50 0.45 0.41/0.50 64 54/87 24 0/73 55 48/71 65 0/96

HSGPA Female 50 0.52 0.49/0.55 67 60/80 17 0/60 64 51/79 51 2/97

(3.39) Male 50 0.48 0.45/0.51 68 58/84 28 1/69 59 50/71 64 3/97

ACTC & Female 50 0.52 0.49/0.55 67 60/83 18 0/65 65 59/78 54 5/95

HSGPA

Male 50 0.47 0.44/0.50 69 59/87 29 2/73 60 54/70 65 9/95

Progress to degree year 4

ACTC Female 48 0.55 0.49/0.58 64 56/82 17 0/63 63 58/75 70 0/99

(24) Male 48 0.46 0.42/0.50 66 54/85 29 0/70 53 46/66 72 0/97

HSGPA Female 49 0.51 0.50/0.56 67 59/79 19 0/57 62 50/79 55 3/97

(3.46) Male 49 0.48 0.46/0.52 68 57/83 30 1/66 57 48/71 68 5/97

ACTC & Female 50 0.52 0.49/0.55 67 59/82 20 0/63 63 57/78 58 4/96

HSGPA

Male 50 0.47 0.45/0.50 69 58/85 31 2/69 58 51/70 69 8/96

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 61 and 50 institutions with available outcomes data for bachelor’s degree completion and

progress to degree analyses, respectively. Med = Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point

average.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

101

Table D-10

Results for Associate’s Degree Completion and Progress to Degree at Two-Year Institutions based on ACTC Score and HSGPA

Models by Gender at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

ACTC Female 25 0.56 0.42/0.63 80 69/92 60 36/84 59 44/65 100 92/100

(29) Male 25 0.44 0.38/0.55 83 61/95 65 21/90 48 35/57 99 91/100

HSGPA Female 5 0.52 0.49/0.52 70 69/72 38 36/44 53 52/55 88 80/92

(3.92) Male 5 0.48 0.47/0.53 75 69/76 50 35/53 50 49/56 92 86/96

ACTC &

HSGPA

Female 25 0.52 0.39/0.56 81 69/92 62 37/83 56 41/62 99 80/100

Male 25 0.46 0.23/0.56 82 61/95 65 22/90 49 41/58 99 85/100

Associate’s degree completion or transfer to four-year institution by year 3

ACTC Female 38 0.54 0.42/0.61 76 63/88 53 21/76 59 46/66 99 76/100

(27) Male 38 0.47 0.43/0.52 77 59/89 55 17/77 52 40/59 97 76/100

HSGPA Female 14 0.49 0.41/0.52 69 65/75 37 23/50 54 44/60 83 58/96

(3.75) Male 14 0.52 0.48/0.56 70 60/76 38 18/51 56 50/62 90 70/98

ACTC & Female 38 0.51 0.41/0.54 77 65/88 53 24/76 56 46/62 97 61/100

HSGPA

Male 38 0.49 0.41/0.55 78 60/89 55 19/78 54 42/63 97 71/100

Progress to degree year 1

ACTC Female 41 0.54 0.36/0.62 66 59/76 14 0/52 67 62/74 53 0/99

(19) Male 41 0.50 0.39/0.56 65 55/84 16 0/67 62 39/73 54 0/99

HSGPA Female 41 0.50 0.32/0.52 66 62/76 14 0/44 65 56/76 44 1/89

(3.03) Male 41 0.51 0.46/0.57 65 59/77 16 0/54 63 52/78 56 2/91

ACTC & Female 42 0.50 0.20/0.59 69 63/77 17 0/52 68 59/76 48 0/99

HSGPA

Male 42 0.50 0.37/0.56 67 59/83 18 0/66 66 36/78 54 0/100

102

Table D-10 (cont.)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

ACTC Female 41 0.54 0.39/0.58 65 53/77 24 2/53 63 51/68 76 14/99

(21) Male 41 0.48 0.39/0.56 65 59/81 27 -1/63 56 45/66 74 17/96

HSGPA Female 41 0.50 0.34/0.53 66 54/75 25 6/51 60 52/67 67 13/97

(3.38) Male 41 0.50 0.42/0.56 66 62/83 29 0/66 58 45/69 76 23/97

ACTC & Female 41 0.51 0.35/0.54 67 55/78 27 6/53 63 52/68 67 15/94

HSGPA

Male 41 0.49 0.41/0.56 67 62/83 30 0/66 59 51/70 73 24/92

Progress to degree year 3

ACTC Female 41 0.54 0.41/0.62 67 57/80 31 7/60 61 49/69 89 35/100

(23) Male 41 0.47 0.40/0.56 69 57/86 37 0/72 53 46/62 87 39/99

HSGPA Female 36 0.50 0.36/0.53 67 55/76 32 11/52 57 48/65 80 31/98

(3.62) Male 36 0.50 0.42/0.56 69 59/82 36 2/64 55 44/68 85 45/99

ACTC & Female 41 0.51 0.37/0.57 69 57/80 33 11/60 60 49/65 79 34/99

HSGPA

Male 41 0.48 0.39/0.56 70 59/86 39 3/73 56 47/68 85 46/98

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. There were 43 and 42 institutions with available outcomes data for associate’s degree completion and

progress to degree analyses, respectively. There were 40 institutions with data available for associate’s degree or transfer to a four-year institution by year 3. Med

= Median; Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

103

Table D-11

Results for Achieving Levels of Year 6 College Cumulative GPA at Four-Year Institutions based on ACTC Score and HSGPA Models

by Gender at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC Female 57 0.60 0.55/0.65 68 64/78 7 1/23 75 70/83 39 18/83

(20) Male 57 0.40 0.37/0.44 65 60/75 20 7/50 57 50/65 44 17/87

HSGPA Female 57 0.56 0.53/0.59 71 66/79 9 1/27 75 63/83 34 17/74

(3.17) Male 57 0.43 0.39/0.44 69 62/75 26 10/48 62 52/68 49 25/84

ACTC &

HSGPA

Female 57 0.56 0.54/0.59 72 66/79 13 3/30 77 68/84 41 21/74

Male 57 0.40 0.36/0.43 71 63/78 29 10/53 61 52/69 53 24/84

3.50 or higher

ACTC Female 57 0.62 0.57/0.70 76 65/90 48 23/79 70 60/75 93 79/100

(27) Male 57 0.39 0.34/0.47 84 71/94 69 46/89 47 40/57 92 72/99

HSGPA Female 40 0.54 0.52/0.58 76 65/84 47 19/68 61 54/69 83 63/96

(3.86) Male 40 0.41 0.38/0.46 84 70/91 68 41/81 47 40/53 89 68/98

ACTC & Female 57 0.57 0.53/0.61 80 66/91 55 18/81 67 59/73 88 67/99

HSGPA

Male 57 0.39 0.32/0.43 87 71/95 73 41/89 51 45/59 91 69/99

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who

graduated with a bachelor’s degree before the end of year 6. There were 57 institutions with data available for the year 6 college GPA analyses. Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

104

Table D-12

Results for Achieving Levels of Year 3 College Cumulative GPA at Two-Year Institutions based on ACTC Score and HSGPA Models

by Gender at Total-Group Optimal Selection Values (SV)

Predictor

variable

(median

SV) Gender

Group-specific

probability of

success

Accuracy rate

(AR)

Increase in AR

(∆AR)

Success rate

(SR)

Observed

percentage below

Med Min/Max Med Min/Max Med Min/Max Med Min/Max Med Min/Max

3.00 or higher

ACTC Female 42 0.58 0.49/0.63 65 62/69 18 5/36 68 62/78 66 33/92

(20) Male 42 0.43 0.39/0.55 66 61/72 30 3/45 53 48/65 67 34/91

HSGPA Female 42 0.53 0.42/0.56 67 61/70 20 8/35 66 59/74 61 24/83

(3.30) Male 42 0.45 0.42/0.53 69 62/76 34 5/51 56 52/65 70 37/89

ACTC &

HSGPA

Female 42 0.53 0.43/0.56 69 63/72 22 10/39 69 62/77 62 29/81

Male 42 0.44 0.40/0.53 71 63/77 37 6/53 58 52/66 69 39/89

3.50 or higher

ACTC Female 42 0.59 0.50/0.66 82 72/88 63 40/77 67 56/75 97 84/100

(26) Male 42 0.41 0.35/0.48 86 75/92 72 50/84 49 37/56 96 84/100

HSGPA Female 5 0.53 0.48/0.54 77 73/78 54 44/55 55 48/56 91 86/94

(3.98) Male 5 0.46 0.43/0.53 80 77/85 61 53/72 49 44/53 95 90/97

ACTC & Female 42 0.54 0.48/0.59 82 73/89 64 42/77 63 56/69 96 82/100

HSGPA

Male 42 0.41 0.33/0.51 87 77/93 75 53/86 50 39/58 96 87/99

Note. All statistics presented in the table are evaluated at the institution-specific total-group optimal selection values that were associated with the maximum

ARs. Total-group optimal selection values (SV) varied substantially across institutions (see Radunzel & Noble, 2012a); median SVs are shown in table. K =

number of institutions with viable total-group models. Students’ cumulative GPAs at degree completion were included in year 3 GPA analyses for students who

graduated with an associate’s degree before the end of year 3. There were 42 institutions with data available for the year 3 college GPA analyses. Med = Median;

Min = Minimum; Max = Maximum; ACTC = ACT Composite; HSGPA = high school grade point average.

Multiple combinations of ACTC score and HSGPA corresponded to a probability of 0.50 for the total-group joint models.

105

Appendix E

Tables E-1 to E-3

106

107

Table E-1

Distributions of Percentages of Students Meeting ACT Benchmarks across Institutions by

Applicant/Enrollment Status, Type of Institution, and Race/Ethnicity

Institution

type

ACT

Benchmark Race/ethnicity

Applicant pool Enrolled students

Med Min/Max Med Min/Max

Four-year

(n = 61)

English

Minority 47 30/66 58 20/100

White 75 56/88 83 39/97

Mathematics

Minority 14 3/32 17 0/67

White 35 16/66 42 4/91

Reading

Minority 31 13/48 37 11/70

White 57 39/74 64 26/90

Science

Minority 7 0/20 11 0/35

White 25 11/46 31 9/63

Two-year

(n = 43)

English

Minority 36 20/51 33 20/83

White 58 47/68 58 46/72

Mathematics

Minority 7 1/20 6 1/31

White 15 8/43 15 6/42

Reading

Minority 22 6/50 22 5/44

White 36 25/53 36 24/53

Science

Minority 4 0/11 4 0/15

White 11 5/22 11 3/24

Note. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. Underrepresented minority students include African American, American Indian, and Hispanic

students. For the typical numbers of students per institution see those reported in Tables A-1 and A-2. Med =

median; Min = minimum; Max = maximum.

108

Table E-2

Distributions of Percentages of Students Meeting ACT Benchmarks across Institutions by

Applicant/Enrollment Status, Type of Institution, and Family Income

Institution

type

ACT

Benchmark Family income

Applicant pool Enrolled students

Med Min/Max Med Min/Max

Four-year

(n = 61)

English

Low

59 26/74 72 24/92

Mid

71 37/83 81 39/95

High

78 47/89 83 38/97

Mathematics

Low

21 3/45 30 4/71

Mid

32 6/56 39 6/83

High

41 11/69 43 0/90

Reading

Low

41 13/61 51 12/75

Mid

52 19/70 62 16/83

High

56 25/75 64 21/89

Science

Low

14 2/34 21 2/51

Mid

21 3/42 27 3/54

High

26 4/50 31 0/62

Two-year

(n = 43)

English

Low

44 29/64 43 31/74

Mid

56 42/69 56 39/71

High

60 47/73 59 44/79

Mathematics

Low

10 3/37 10 3/38

Mid

14 8/39 14 6/37

High

19 10/40 18 4/42

Reading

Low

29 14/47 28 15/53

Mid

34 23/48 34 19/47

High

39 26/52 38 23/59

Science

Low

8 2/16 7 2/17

Mid

10 4/25 10 2/28

High

13 3/23 11 2/22

Note. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. Low is for lower-income students (annual family income < $30,000), Mid is for middle-income

students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000). For the typical numbers of students per institution see those reported in Tables A-3 and

A-4. Med = median; Min = minimum; Max = maximum.

109

Table E-3

Distributions of Percentages of Students Meeting ACT Benchmarks across Institutions by

Applicant/Enrollment Status, Type of Institution, and Gender

Institution

type

ACT

Benchmark Gender

Applicant pool Enrolled students

Med Min/Max Med Min/Max

Four-year

(n = 61)

English

Female

72 34/87 82 39/97

Male

65 28/82 75 22/94

Mathematics

Female

28 5/53 36 5/82

Male

34 5/66 43 5/90

Reading

Female

53 18/71 62 18/85

Male

49 14/68 56 11/83

Science

Female

18 2/39 24 2/48

Male

25 3/51 31 2/65

Two-year

(n = 43)

English

Female

55 39/70 56 40/76

Male

48 35/62 48 31/67

Mathematics

Female

11 5/38 11 5/36

Male

16 9/39 15 4/41

Reading

Female

35 19/51 34 18/55

Male

31 18/42 30 17/44

Science

Female

8 3/18 7 1/20

Male

12 5/25 12 3/27

Note. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. For the typical numbers of students per institution see those reported in Tables A-5 and A-6. Med =

median; Min = minimum; Max = maximum.

110

111

Appendix F

Tables F-1 to F-6

112

113

Table F-1

Results for Bachelor’s Degree Completion, Progress to Degree, and Achieving Levels of Year 6

College Cumulative GPA at Four-Year Institutions based on ACT College Readiness

Benchmarks by Race/Ethnicity

Subject area

(median

total-group

probability

of success)

Race/

ethnicity

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

English White 0.34 0.12/0.77 46 22/80 4 1/7

(0.35) Minority 0.28 0.10/0.64 36 16/70 8 4/13

Mathematics White 0.47 0.23/0.79 53 30/82 11 2/24

(0.46) Minority 0.42 0.21/0.69 48 29/74 17 7/31

Reading White 0.41 0.18/0.79 48 23/81 4 1/9

(0.39) Minority 0.32 0.14/0.68 39 18/72 8 4/16

Science White 0.48 0.24/0.80 52 29/81 9 1/17

(0.47) Minority 0.42 0.22/0.72 45 26/74 14 5/25

Progress to degree year 1

English White 0.60 0.29/0.84 73 48/89 5 1/14

(0.59) Minority 0.53 0.17/0.80 61 29/85 13 6/22

Mathematics White 0.74 0.55/0.87 82 67/90 13 3/39

(0.73) Minority 0.71 0.35/0.86 79 46/89 26 9/53

Reading White 0.69 0.40/0.87 75 50/90 5 1/16

(0.67) Minority 0.61 0.24/0.84 67 33/87 14 4/26

Science White 0.77 0.55/0.90 81 61/91 10 2/32

(0.76) Minority 0.72 0.41/0.87 76 47/89 23 6/43

Progress to degree year 2

English White 0.47 0.17/0.75 60 34/82 6 1/10

(0.45) Minority 0.39 0.14/0.72 50 23/78 11 7/18

Mathematics White 0.61 0.38/0.78 70 49/84 13 4/31

(0.61) Minority 0.58 0.28/0.77 65 37/82 25 11/40

Reading White 0.54 0.27/0.80 62 36/84 6 2/12

(0.53) Minority 0.44 0.18/0.77 51 24/81 11 6/19

Science White 0.63 0.39/0.83 69 47/84 12 2/26

(0.62) Minority 0.60 0.31/0.81 64 37/83 24 7/37

114

Table F-1 (cont.)

Subject area

(total-group

probability

of success)

Race/

ethnicity

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

English White 0.40 0.13/0.73 52 28/80 5 2/9

(0.39) Minority 0.33 0.11/0.69 42 18/76 10 6/14

Mathematics White 0.54 0.31/0.76 62 43/81 14 3/27

(0.53) Minority 0.49 0.23/0.74 57 31/80 22 11/36

Reading White 0.47 0.22/0.78 54 29/81 6 2/10

(0.46) Minority 0.37 0.15/0.74 44 20/78 10 6/18

Science White 0.56 0.32/0.80 61 39/82 13 2/23

(0.56) Minority 0.50 0.25/0.79 55 30/81 21 8/34

Progress to degree year 4

English White 0.38 0.13/0.74 49 26/80 5 2/8

(0.37) Minority 0.31 0.11/0.67 39 17/74 9 6/13

Mathematics White 0.51 0.28/0.76 58 38/81 13 3/23

(0.51) Minority 0.45 0.21/0.73 53 28/78 21 10/33

Reading White 0.44 0.21/0.78 52 27/81 5 2/10

(0.43) Minority 0.35 0.14/0.72 41 18/76 10 6/17

Science White 0.53 0.30/0.80 58 36/82 11 2/20

(0.53) Minority 0.47 0.22/0.76 51 27/79 18 7/31

Year 6 cumulative GPA 3.00 or higher

English White 0.50 0.34/0.66 68 51/79 7 4/11

(0.46) Minority 0.34 0.20/0.62 47 34/69 14 10/20

Mathematics White 0.68 0.50/0.79 76 60/86 14 6/24

(0.66) Minority 0.55 0.38/0.75 65 48/80 28 19/42

Reading White 0.63 0.45/0.73 72 53/82 9 5/13

(0.59) Minority 0.43 0.32/0.68 54 41/73 18 13/24

Science White 0.73 0.54/0.82 77 60/86 14 7/21

(0.72) Minority 0.60 0.46/0.78 66 51/80 30 21/40

Note. These analyses were based on all institutions with available data for each outcome (61 institutions for

bachelor’s degree completion, 50 for progress to degree outcomes, and 57 institutions for year 6 cumulative GPA).

Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who graduated

with a bachelor’s degree before the end of year 6. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in

English, mathematics, reading, and science, respectively. Underrepresented minority students include African

American, American Indian, and Hispanic students. Med = Median; Min = Minimum; Max = Maximum.

115

Table F-2

Results for Associate’s Degree Completion, Progress to Degree, and Achieving Levels of Year 3

College Cumulative GPA at Two-Year Institutions based on ACT College Readiness Benchmarks

by Race/Ethnicity

Subject area

(total-group

probability

of success)

Race/

ethnicity

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

English White 0.13 0.04/0.36 19 6/42 4 1/8

(0.12) Minority 0.10 0.03/0.31 15 4/40 6 2/12

Mathematics White 0.22 0.07/0.51 29 10/59 13 3/23

(0.22) Minority 0.19 0.06/0.50 24 8/59 16 5/31

Reading White 0.16 0.05/0.38 19 7/43 5 2/8

(0.15) Minority 0.12 0.04/0.34 15 5/41 6 2/13

Science White 0.23 0.08/0.49 26 9/53 11 4/18

(0.23) Minority 0.20 0.06/0.48 23 7/53 13 4/25

Associate’s degree completion or transfer to four-year institution by year 3

English White 0.22 0.10/0.43 30 15/52 5 3/10

(0.21) Minority 0.18 0.08/0.39 25 11/49 9 4/15

Mathematics White 0.36 0.17/0.62 42 21/70 18 9/26

(0.36) Minority 0.31 0.13/0.60 37 17/68 21 11/33

Reading White 0.26 0.12/0.47 31 15/53 6 3/10

(0.25) Minority 0.21 0.08/0.42 25 11/49 9 4/14

Science White 0.36 0.16/0.59 39 18/63 14 8/19

(0.35) Minority 0.30 0.12/0.55 33 15/61 18 8/25

Progress to degree year 1

English White 0.51 0.17/0.78 63 28/82 10 4/14

(0.49) Minority 0.44 0.11/0.79 55 16/84 18 5/25

Mathematics White 0.71 0.24/0.86 78 30/89 24 8/37

(0.70) Minority 0.68 0.18/0.89 75 31/92 36 17/48

Reading White 0.58 0.22/0.80 65 29/83 11 4/18

(0.56) Minority 0.50 0.13/0.80 56 21/84 19 8/29

Science White 0.70 0.31/0.85 74 39/87 20 10/31

(0.70) Minority 0.65 0.21/0.88 69 30/90 31 15/46

116

Table F-2 (cont.)

Subject area

(total-group

probability

of success)

Race/

ethnicity

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

English White 0.40 0.11/0.63 50 16/70 8 2/11

(0.38) Minority 0.33 0.08/0.60 42 11/68 13 3/19

Mathematics White 0.57 0.14/0.78 63 17/82 21 4/28

(0.57) Minority 0.53 0.12/0.78 60 19/83 30 9/39

Reading White 0.45 0.13/0.66 51 16/71 9 3/12

(0.44) Minority 0.37 0.09/0.62 43 13/69 14 6/19

Science White 0.57 0.17/0.75 61 20/78 18 7/24

(0.57) Minority 0.51 0.13/0.75 55 17/79 26 9/35

Progress to degree year 3

English White 0.33 0.06/0.56 42 10/64 7 2/9

(0.32) Minority 0.28 0.04/0.53 36 5/61 12 2/16

Mathematics White 0.49 0.09/0.71 55 11/77 20 3/24

(0.49) Minority 0.45 0.07/0.71 52 12/78 27 7/34

Reading White 0.37 0.08/0.59 43 11/64 8 2/9

(0.37) Minority 0.31 0.05/0.56 37 8/62 13 4/17

Science White 0.49 0.10/0.69 52 12/73 16 4/21

(0.49) Minority 0.44 0.07/0.69 47 9/73 23 5/29

Year 3 cumulative GPA 3.00 or higher

English White 0.44 0.33/0.62 57 45/73 10 7/15

(0.42) Minority 0.37 0.24/0.56 47 31/67 17 12/20

Mathematics White 0.63 0.50/0.82 70 57/87 22 14/29

(0.63) Minority 0.59 0.42/0.80 65 47/87 35 27/41

Reading White 0.51 0.41/0.68 61 48/77 13 8/19

(0.50) Minority 0.43 0.29/0.60 52 33/69 20 9/25

Science White 0.64 0.52/0.80 69 56/84 20 15/27

(0.63) Minority 0.58 0.41/0.76 63 45/80 31 24/44

Note. These analyses were based on all institutions with available data for each outcome (43 institutions for

associate’s degree completion, 40 for associate’s degree completion or transfer to a four-year institution, and 42 for

progress to degree outcomes and year 3 cumulative GPA). Students’ cumulative GPAs at degree completion were

included in year 3 GPA analyses for students who graduated with an associate’s degree before the end of year 3. The

ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. Underrepresented minority students include African American, American Indian, and Hispanic

students. Med = Median; Min = Minimum; Max = Maximum.

117

Table F-3

Results for Bachelor’s Degree Completion, Progress to Degree, and Achieving Levels of Year 6

College Cumulative GPA at Four-Year Institutions based on ACT College Readiness

Benchmarks by Family Income

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

English Low 0.29 0.09/0.73 38 15/76 6 2/11

(0.35) Mid 0.35 0.12/0.75 44 22/79 5 1/12

High 0.39 0.15/0.78 50 27/82 4 1/10

Mathematics Low 0.41 0.17/0.75 48 23/79 14 5/30

(0.46) Mid 0.45 0.23/0.78 52 30/81 12 3/28

High 0.50 0.27/0.81 57 34/84 10 2/25

Reading Low 0.34 0.12/0.74 39 15/76 6 1/12

(0.39) Mid 0.41 0.18/0.77 46 23/79 5 1/12

High 0.45 0.22/0.81 52 27/83 4 1/13

Science Low 0.42 0.17/0.76 46 20/77 11 3/23

(0.47) Mid 0.48 0.25/0.79 51 29/80 10 2/23

High 0.52 0.29/0.82 57 34/83 9 1/23

Progress to degree year 1

English Low 0.54 0.19/0.81 63 33/84 8 3/19

(0.59) Mid 0.58 0.28/0.84 71 46/87 6 2/20

High 0.62 0.30/0.86 76 52/90 5 1/17

Mathematics Low 0.68 0.46/0.84 77 59/87 18 6/52

(0.73) Mid 0.74 0.53/0.87 82 66/90 14 4/43

High 0.77 0.55/0.89 85 69/92 11 2/37

Reading Low 0.60 0.28/0.83 66 36/85 8 2/20

(0.67) Mid 0.68 0.37/0.85 74 48/89 6 2/20

High 0.72 0.41/0.88 78 53/91 5 1/19

Science Low 0.69 0.44/0.85 74 51/87 17 5/42

(0.76) Mid 0.77 0.55/0.88 81 63/90 12 3/39

High 0.80 0.58/0.91 84 66/92 9 2/36

118

Table F-3 (cont.)

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

English Low 0.38 0.12/0.71 51 22/76 7 4/14

(0.45) Mid 0.46 0.17/0.73 58 33/80 6 3/14

High 0.50 0.19/0.77 63 39/84 5 1/13

Mathematics Low 0.54 0.30/0.74 63 41/79 18 8/38

(0.61) Mid 0.62 0.37/0.76 69 49/82 15 5/35

High 0.66 0.40/0.80 73 54/85 12 4/32

Reading Low 0.44 0.18/0.75 52 25/78 7 3/15

(0.53) Mid 0.54 0.26/0.78 60 35/82 6 2/14

High 0.59 0.29/0.82 66 40/85 6 2/15

Science Low 0.55 0.31/0.79 60 37/80 16 5/31

(0.62) Mid 0.63 0.40/0.81 69 49/83 13 4/31

High 0.68 0.43/0.85 73 53/86 11 2/30

Progress to degree year 3

English Low 0.33 0.08/0.68 44 17/73 7 3/13

(0.39) Mid 0.39 0.13/0.72 51 26/78 6 3/13

High 0.43 0.15/0.75 56 33/81 5 1/12

Mathematics Low 0.47 0.23/0.71 56 33/76 17 6/32

(0.53) Mid 0.53 0.30/0.75 63 42/80 15 4/31

High 0.57 0.34/0.78 66 48/83 12 3/28

Reading Low 0.38 0.13/0.72 45 19/75 7 3/15

(0.46) Mid 0.47 0.20/0.76 53 29/79 6 3/12

High 0.52 0.24/0.79 59 34/83 6 2/13

Science Low 0.49 0.23/0.75 53 28/77 16 5/28

(0.56) Mid 0.56 0.32/0.79 62 40/81 13 3/27

High 0.61 0.37/0.82 67 46/84 12 2/28

119

Table F-3 (cont.)

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 4

English Low 0.31 0.09/0.68 41 16/73 6 3/12

(0.37) Mid 0.37 0.13/0.72 49 25/78 6 2/11

High 0.41 0.15/0.75 54 31/81 5 1/10

Mathematics Low 0.44 0.21/0.71 51 30/76 16 5/27

(0.51) Mid 0.50 0.28/0.75 58 38/80 14 4/26

High 0.54 0.32/0.78 62 43/83 12 3/24

Reading Low 0.36 0.13/0.71 42 18/74 7 3/13

(0.43) Mid 0.43 0.19/0.76 51 27/79 6 2/11

High 0.49 0.23/0.80 56 32/82 5 2/12

Science Low 0.44 0.22/0.74 50 26/76 14 5/25

(0.53) Mid 0.52 0.30/0.78 58 37/80 12 3/23

High 0.57 0.34/0.82 63 42/83 11 2/24

Year 6 cumulative GPA 3.00 or higher

English Low 0.42 0.29/0.64 60 44/74 12 8/20

(0.46) Mid 0.46 0.33/0.61 65 49/78 9 5/17

High 0.48 0.35/0.63 68 52/80 7 4/14

Mathematics Low 0.64 0.44/0.76 73 54/85 24 12/42

(0.66) Mid 0.66 0.48/0.78 75 60/86 17 10/31

High 0.68 0.50/0.79 77 61/87 14 7/26

Reading Low 0.54 0.40/0.67 65 50/76 15 9/24

(0.59) Mid 0.59 0.43/0.71 70 52/81 11 7/19

High 0.62 0.45/0.73 73 54/83 9 5/17

Science Low 0.68 0.50/0.80 74 55/85 24 14/39

(0.72) Mid 0.71 0.54/0.81 76 60/86 18 12/30

High 0.73 0.55/0.83 78 61/87 15 8/24

Note. These analyses were based on all institutions with available data for each outcome (61 institutions for

bachelor’s degree completion, 50 for progress to degree outcomes, and 57 institutions for year 6 cumulative GPA).

Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who graduated

with a bachelor’s degree before the end of year 6. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in

English, mathematics, reading, and science, respectively. Low is for lower-income students (annual family income <

$30,000), Mid is for middle-income students (annual family income between $30,000 and $60,000), and High is for

higher-income students (annual family income > $60,000). Med = Median; Min = Minimum; Max = Maximum.

120

Table F-4

Results for Associate’s Degree Completion, Progress to Degree, and Achieving Levels of Year 3

College Cumulative GPA at Two-Year Institutions based on ACT College Readiness Benchmarks

by Family Income

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

English Low 0.11 0.04/0.30 17 6/36 5 2/10

(0.12) Mid 0.13 0.04/0.37 20 6/44 4 1/8

High 0.13 0.04/0.37 20 7/43 3 1/7

Mathematics Low 0.21 0.07/0.47 27 10/55 15 4/27

(0.22) Mid 0.24 0.07/0.53 29 10/61 14 4/24

High 0.22 0.07/0.49 27 9/57 11 3/20

Reading Low 0.13 0.04/0.32 18 6/38 5 2/11

(0.15) Mid 0.17 0.05/0.39 21 7/45 5 2/9

High 0.17 0.05/0.39 21 7/44 4 1/9

Science Low 0.21 0.07/0.45 25 9/49 12 4/22

(0.23) Mid 0.24 0.07/0.51 28 9/55 12 3/19

High 0.23 0.07/0.49 27 9/54 10 3/17

Associate’s degree completion or transfer to four-year institution by year 3

English Low 0.18 0.08/0.36 25 11/44 6 3/11

(0.21) Mid 0.23 0.09/0.44 31 13/54 6 3/10

High 0.26 0.12/0.47 35 17/56 5 3/9

Mathematics Low 0.31 0.13/0.55 37 17/64 19 10/30

(0.36) Mid 0.38 0.15/0.64 45 19/72 19 10/26

High 0.40 0.18/0.63 45 23/70 16 9/22

Reading Low 0.21 0.08/0.39 25 10/46 7 4/12

(0.25) Mid 0.27 0.10/0.48 32 13/55 7 3/11

High 0.31 0.13/0.52 36 16/58 7 3/10

Science Low 0.30 0.12/0.52 34 14/56 15 7/22

(0.35) Mid 0.37 0.14/0.61 41 16/65 15 7/21

High 0.41 0.17/0.62 44 21/66 14 8/18

121

Table F-4 (cont.)

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 1

English Low 0.44 0.14/0.75 57 23/79 14 5/22

(0.49) Mid 0.51 0.16/0.80 63 27/84 11 5/18

High 0.55 0.18/0.83 67 27/86 9 3/16

Mathematics Low 0.67 0.22/0.85 75 29/88 31 10/48

(0.70) Mid 0.72 0.23/0.87 79 30/90 25 9/40

High 0.73 0.24/0.88 79 28/90 19 7/35

Reading Low 0.50 0.18/0.76 59 24/80 15 6/28

(0.56) Mid 0.57 0.21/0.81 65 28/84 12 4/21

High 0.62 0.23/0.84 70 27/86 11 3/20

Science Low 0.66 0.27/0.84 70 33/85 26 13/38

(0.70) Mid 0.71 0.30/0.86 75 36/88 21 10/37

High 0.74 0.31/0.88 78 42/90 18 7/33

Progress to degree year 2

English Low 0.34 0.09/0.57 43 13/64 10 2/16

(0.38) Mid 0.41 0.10/0.64 51 14/71 8 2/13

High 0.45 0.12/0.68 56 16/75 7 2/12

Mathematics Low 0.53 0.12/0.74 60 15/80 26 5/36

(0.57) Mid 0.58 0.12/0.79 65 16/84 22 4/30

High 0.60 0.15/0.80 66 17/84 19 3/28

Reading Low 0.39 0.10/0.59 45 13/65 11 3/18

(0.44) Mid 0.46 0.11/0.67 52 15/72 9 3/14

High 0.51 0.14/0.70 56 16/75 8 2/14

Science Low 0.52 0.14/0.71 57 17/74 21 7/30

(0.57) Mid 0.59 0.15/0.77 63 18/79 19 6/28

High 0.62 0.17/0.78 66 22/81 17 8/25

122

Table F-4 (cont.)

Subject area

(total-group

probability

of success)

Family

income

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

English Low 0.29 0.05/0.50 37 8/58 9 2/13

(0.32) Mid 0.34 0.05/0.58 43 9/65 8 2/11

High 0.38 0.07/0.61 48 10/68 6 2/9

Mathematics Low 0.45 0.08/0.67 53 10/74 23 4/31

(0.49) Mid 0.51 0.08/0.73 58 10/78 21 3/28

High 0.52 0.09/0.73 58 11/78 17 3/24

Reading Low 0.33 0.06/0.53 39 9/59 10 2/13

(0.37) Mid 0.38 0.07/0.61 44 10/66 9 2/11

High 0.42 0.09/0.64 48 10/69 7 2/10

Science Low 0.44 0.09/0.65 48 10/69 19 3/26

(0.49) Mid 0.50 0.09/0.71 54 10/74 18 3/25

High 0.54 0.10/0.73 57 12/76 15 4/22

Year 3 cumulative GPA 3.00 or higher

English Low 0.42 0.34/0.60 53 45/67 14 9/17

(0.42) Mid 0.43 0.34/0.59 57 45/72 12 8/15

High 0.41 0.29/0.58 56 41/72 10 7/15

Mathematics Low 0.64 0.53/0.79 70 59/85 29 19/36

(0.63) Mid 0.64 0.53/0.80 71 60/86 24 17/30

High 0.60 0.47/0.79 67 54/86 21 15/28

Reading Low 0.48 0.41/0.62 58 49/71 17 10/24

(0.50) Mid 0.52 0.42/0.67 62 49/75 15 10/20

High 0.49 0.37/0.65 59 44/75 13 8/19

Science Low 0.63 0.54/0.76 68 58/80 27 22/33

(0.63) Mid 0.65 0.54/0.79 70 58/83 24 18/31

High 0.60 0.47/0.77 66 50/82 20 13/27

Note. These analyses were based on all institutions with available data for each outcome (43 institutions for

associate’s degree completion, 40 for associate’s degree completion or transfer to a four-year institution, and 42 for

progress to degree outcomes and year 3 cumulative GPA). Students’ cumulative GPAs at degree completion were

included in year 3 GPA analyses for students who graduated with an associate’s degree before the end of year 3. The

ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. Low is for lower-income students (annual family income < $30,000), Mid is for middle-income

students (annual family income between $30,000 and $60,000), and High is for higher-income students (annual

family income > $60,000). Med = Median; Min = Minimum; Max = Maximum.

123

Table F-5

Results for Bachelor’s Degree Completion, Progress to Degree, and Achieving Levels of Year 6

College Cumulative GPA at Four-Year Institutions based on ACT College Readiness

Benchmarks by Gender

Subject area

(total-group

probability

of success) Gender

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Bachelor’s degree completion by year 6

English Female 0.37 0.12/0.77 48 23/81 5 1/11

(0.35) Male 0.32 0.11/0.74 41 18/77 5 1/11

Mathematics Female 0.52 0.28/0.81 59 36/84 14 5/32

(0.46) Male 0.39 0.18/0.75 47 25/78 11 3/28

Reading Female 0.43 0.18/0.80 50 25/82 5 1/13

(0.39) Male 0.36 0.15/0.76 42 19/78 5 1/12

Science Female 0.54 0.29/0.82 58 34/84 13 3/27

(0.47) Male 0.41 0.19/0.77 46 24/78 10 2/21

Progress to degree year 1

English Female 0.63 0.25/0.86 74 45/90 6 1/20

(0.59) Male 0.54 0.25/0.81 66 41/85 7 2/22

Mathematics Female 0.80 0.60/0.90 87 73/93 16 4/49

(0.73) Male 0.67 0.42/0.83 77 58/87 14 3/44

Reading Female 0.71 0.37/0.88 77 48/91 6 2/21

(0.67) Male 0.62 0.32/0.82 69 42/87 7 2/20

Science Female 0.83 0.60/0.92 86 67/93 14 4/47

(0.76) Male 0.70 0.45/0.85 75 54/88 13 3/39

Progress to degree year 2

English Female 0.48 0.16/0.76 62 33/84 6 2/15

(0.45) Male 0.42 0.15/0.70 54 29/78 7 3/16

Mathematics Female 0.67 0.43/0.81 74 55/87 18 6/41

(0.61) Male 0.54 0.29/0.71 63 44/79 14 5/34

Reading Female 0.56 0.26/0.82 64 36/85 6 2/16

(0.53) Male 0.48 0.21/0.76 56 31/80 7 3/15

Science Female 0.69 0.45/0.86 75 53/88 16 5/39

(0.62) Male 0.56 0.32/0.78 63 41/81 13 4/30

124

Table F-5 (cont.)

Subject area

(total-group

probability

of success) Gender

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 3

English Female 0.42 0.12/0.77 53 26/82 6 2/13

(0.39) Male 0.35 0.11/0.69 46 23/75 6 3/13

Mathematics Female 0.59 0.35/0.80 68 48/85 18 5/36

(0.53) Male 0.47 0.24/0.69 56 37/76 14 4/29

Reading Female 0.49 0.20/0.81 57 29/84 6 2/15

(0.46) Male 0.40 0.17/0.73 49 25/76 7 2/13

Science Female 0.62 0.37/0.84 68 45/86 16 5/36

(0.56) Male 0.50 0.26/0.75 55 34/78 13 4/27

Progress to degree year 4

English Female 0.40 0.12/0.77 51 25/82 5 2/12

(0.37) Male 0.33 0.11/0.69 44 22/75 6 3/12

Mathematics Female 0.56 0.32/0.80 63 43/85 17 5/32

(0.51) Male 0.43 0.22/0.69 52 34/76 13 4/25

Reading Female 0.46 0.19/0.81 54 27/83 6 2/14

(0.43) Male 0.38 0.16/0.73 46 24/76 7 2/12

Science Female 0.59 0.34/0.83 64 41/85 16 5/29

(0.53) Male 0.46 0.24/0.75 51 31/77 12 4/22

Year 6 cumulative GPA 3.00 or higher

English Female 0.53 0.39/0.69 72 56/84 9 4/19

(0.46) Male 0.39 0.28/0.51 55 40/68 10 5/19

Mathematics Female 0.76 0.59/0.87 84 69/93 20 12/35

(0.66) Male 0.52 0.36/0.65 64 47/78 20 10/35

Reading Female 0.66 0.50/0.78 76 60/87 12 7/22

(0.59) Male 0.49 0.35/0.60 59 43/72 12 7/20

Science Female 0.81 0.66/0.89 85 72/92 20 14/35

(0.72) Male 0.58 0.42/0.71 67 48/78 19 11/32

Note. These analyses were based on all institutions with available data for each outcome (61 institutions for

bachelor’s degree completion, 50 for progress to degree outcomes, and 57 institutions for year 6 cumulative GPA).

Students’ cumulative GPAs at degree completion were included in year 6 GPA analyses for students who graduated

with a bachelor’s degree before the end of year 6. The ACT College Readiness Benchmarks are 18, 22, 21, and 24 in

English, mathematics, reading, and science, respectively. Med = Median; Min = Minimum; Max = Maximum.

125

Table F-6

Results for Associate’s Degree Completion, Progress to Degree, and Achieving Levels of Year 3

College Cumulative GPA at Two-Year Institutions based on ACT College Readiness Benchmarks

by Gender

Subject area

(total-group

probability

of success) Gender

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Associate’s degree completion by year 3

English Female 0.13 0.04/0.32 20 7/40 4 1/9

(0.12) Male 0.11 0.03/0.40 16 5/44 4 1/9

Mathematics Female 0.27 0.09/0.53 34 13/61 19 5/28

(0.22) Male 0.18 0.05/0.49 24 6/57 11 2/23

Reading Female 0.17 0.05/0.36 21 8/43 6 2/10

(0.15) Male 0.13 0.04/0.42 17 5/45 5 1/9

Science Female 0.27 0.09/0.51 31 11/55 15 6/23

(0.23) Male 0.18 0.05/0.48 22 6/52 9 2/18

Associate’s degree completion or transfer to four-year institution by year 3

English Female 0.21 0.10/0.40 30 15/50 6 3/10

(0.21) Male 0.22 0.08/0.44 28 11/52 6 3/11

Mathematics Female 0.39 0.19/0.64 46 25/71 23 13/30

(0.36) Male 0.32 0.12/0.58 39 16/67 16 8/26

Reading Female 0.25 0.11/0.45 31 15/52 8 4/12

(0.25) Male 0.24 0.09/0.46 29 11/52 7 4/11

Science Female 0.38 0.18/0.60 42 21/64 19 11/24

(0.35) Male 0.32 0.12/0.56 35 14/61 13 7/20

Progress to degree year 1

English Female 0.49 0.17/0.77 63 30/82 11 6/22

(0.49) Male 0.49 0.15/0.80 59 21/83 12 4/20

Mathematics Female 0.76 0.29/0.88 82 38/91 30 13/47

(0.70) Male 0.65 0.17/0.86 74 22/88 24 6/40

Reading Female 0.57 0.23/0.79 65 31/83 14 6/27

(0.56) Male 0.53 0.17/0.80 62 23/83 13 4/26

Science Female 0.75 0.37/0.87 78 46/89 25 13/42

(0.70) Male 0.64 0.24/0.85 69 31/87 21 9/34

126

Table F-6 (cont.)

Subject area

(total-group

probability

of success) Gender

Group-specific

probability of

success at

Benchmark

Success rate

(SR)

Increase in SR

(∆SR)

Med Min/Max Med Min/Max Med Min/Max

Progress to degree year 2

English Female 0.38 0.11/0.60 50 16/68 9 3/17

(0.38) Male 0.38 0.09/0.64 46 11/70 9 2/14

Mathematics Female 0.63 0.17/0.80 70 21/84 27 6/37

(0.57) Male 0.50 0.09/0.76 58 10/81 20 2/32

Reading Female 0.45 0.14/0.64 52 18/70 11 3/18

(0.44) Male 0.41 0.09/0.65 47 12/70 9 2/16

Science Female 0.61 0.20/0.77 65 24/79 23 10/33

(0.57) Male 0.51 0.11/0.74 56 14/77 18 5/27

Progress to degree year 3

English Female 0.33 0.07/0.53 43 11/62 8 2/13

(0.32) Male 0.31 0.04/0.58 39 6/65 8 1/11

Mathematics Female 0.55 0.11/0.73 62 15/78 26 5/33

(0.49) Male 0.42 0.05/0.70 50 6/76 19 2/29

Reading Female 0.38 0.09/0.57 45 12/63 10 3/12

(0.37) Male 0.34 0.05/0.60 40 7/64 9 2/11

Science Female 0.53 0.13/0.71 57 15/74 22 6/30

(0.49) Male 0.43 0.06/0.68 47 8/72 16 2/22

Year 3 cumulative GPA 3.00 or higher

English Female 0.46 0.34/0.64 59 45/77 12 8/15

(0.42) Male 0.36 0.27/0.61 47 37/67 12 8/16

Mathematics Female 0.71 0.57/0.90 78 63/93 30 19/35

(0.63) Male 0.50 0.39/0.70 59 46/79 23 16/34

Reading Female 0.55 0.42/0.73 64 49/81 16 10/20

(0.50) Male 0.42 0.33/0.62 51 40/67 15 9/18

Science Female 0.71 0.57/0.88 75 62/90 27 22/34

(0.63) Male 0.53 0.41/0.72 59 44/76 22 16/29

Note. These analyses were based on all institutions with available data for each outcome (43 institutions for

associate’s degree completion, 40 for associate’s degree completion or transfer to a four-year institution, and 42 for

progress to degree outcomes and year 3 cumulative GPA). Students’ cumulative GPAs at degree completion were

included in year 3 GPA analyses for students who graduated with an associate’s degree before the end of year 3. The

ACT College Readiness Benchmarks are 18, 22, 21, and 24 in English, mathematics, reading, and science,

respectively. Med = Median; Min = Minimum; Max = Maximum.

*050205130*

Rev 1

Differential Effects on Student Demographic

Groups of Using ACT

College Readiness

Assessment Composite Score, ACT

Benchmarks, and High School Grade Point

Average for Predicting Long-Term College

Success through Degree Completion

Justine Radunzel

Julie Noble

August 2013

ACT Research

Report Series

2013 (5)