Does Superscoring Increase Subgroup Differences?

ACT Research & Policy | Technical Brief | July 2019

Krista Mattern, PhD, and Justine Radunzel, PhD

When applicants take the ACT

more than once,

how do colleges and universities reconcile and

make sense out of the multiple scores? In terms

of validity, fairness, and impact on subgroup

differences, are certain score-use polices better

than others? Given that the proportion of

students retaking the ACT has increased over

time (Harmston & Crouse, 2016), answers to

these questions have become increasingly

relevant and pressing. The focus of this

technical brief is to summarize empirical

evidence on the validity and fairness of various

score-use policies with an emphasis on

superscoring. Specifically, findings from a study

that examined the differential validity and

predictions of different score-use policies that

was published in 2018 will be reviewed.

Additionally, new analyses demonstrating the

impact of superscoring on subgroup differences

will be presented. Finally, responses to ACT’s

Higher Education Score Use Survey are

presented to help contextualize these findings.

The intent is to arm higher education

professionals with the most recent evidence to

help support informed decision making on their

own campus.

Background

Composite score (superscoring). And still others

have different and sometimes multiple policies in

place. In particular, based on responses from

115 higher education professionals who

completed ACT’s Higher Education Score Use

Survey in April of 2019, 33% indicated that they

superscore the ACT, whereas 49% indicated

that they superscore the SAT (ACT, 2019). The

percentage that superscore both tests was 32%.

Of those that superscore the ACT, all but one

said they also superscore the SAT with the one

exception indicating that they do not use the

SAT. Among those that superscore the SAT, a

different pattern emerges: 66% superscore the

ACT, 32% use the highest score from a single

administration on the ACT, and 2% do an in-

depth review of all the ACT scores. This finding

raises two issues that need further attention.

First, given that superscoring is a fairly common

practice among postsecondary institutions, what

are the implications of this score-use policy in

terms of validity and fairness? Second, given

that many institutions have inconsistent score-

use policies depending on the test (ACT vs.

SAT), what are the validity and fairness

implications for ACT test-takers? This report will

focus on the first issue. ACT has provided the

following recommendations as it relates to the

second issue:

A survey of the current landscape of college

admissions found that there isn’t a “one-size-fits-

all” solution to how institutions of higher

education treat multiple test records from the

same applicant. Some postsecondary

institutions use a student's most recent score.

Others “pick and choose,” selecting the best

scores a student has earned in each content

area over the course of multiple test

administrations and forming a combined highest

1. Con

sistency. Whatever score-use policy an

institution chooses, that policy should be

applied consistently to all applicants.

Concerns of fairness arise if one score-use

policy (most recent score) is applied to some

groups of applicants (e.g., females, ACT test-

takers) and a different score-use policy

(superscore) is applied to other groups of

applicants (e.g., males, SAT test-takers).

ACT.org/research

ACT Research & Policy | Technical Brief | July 2019

2. Concordance. For institutions that receive

both ACT and SAT scores from applicants,

the 2018 ACT-SAT concordance should be

used to convert SAT scores to ACT scores

and vice versa (The College Board & ACT,

2018). Given the change in the score scale

for the 2016 SAT, using the previous ACT-

SAT concordance puts ACT test-takers at an

unfair disadvantage (ACT, 2009).

Validity and Fairness of

Superscoring

ACT Working Paper with an overview provided

in the 2017 Higher Education Research Digest

(Mattern, Radunzel, Bertling, & Ho, 2017; ACT,

2017) and later published in a peer-reviewed

journal (Mattern, Radunzel, Bertling, & Ho,

2018). Contrary to expectations, the results

showed that scores based on the superscoring

method (referred to as superscores) were just as

predictive (actually slightly more predictive) of

first-year grades as compared to other scoring

methods (recent, average, highest). Moreover,

superscoring resulted in the least amount of

differential prediction by the number of times a

ACT has been examining the validity and

student tests. Interestingly, we found that first-

fairness of different scoring practices over the

year grades for students who tested more often

last several years. Results from an initial study

were underpredicted even when prediction

were first made publicly available as both an

models were based on superscores (see Figure

1).

Figure 1. Magnitude of Differential Prediction by Number of Testing Occasions and Four Composite

Scoring Methods when ACT Composite Score is Held Constant at the Sample Mean of 23

-0.20

0.00

0.14

0.26

-0.23

-0.01

0.19

0.32

-0.18

-0.01

0.13

0.23

-0.15

-0.01

0.10

0.19

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

1 Time 2 Times 3 Times 4 or More Times

Prediction Error

Number of Testing Occasions

Last Average Highest Superscore

Note: Prediction error is calculated by subtracting one’s expected FYGPA based on the overall model from the

expected value based on the model that includes retesting subgroup indicators and the interaction between the

ACT Composite score and retesting indicators. Negative values indicate overprediction; positive values indicate

underprediction.

ACT Research & Policy | Technical Brief | July 2019

As shown in Figure 1, retesters performed better

in college than what was expected based on

their test scores. And this prediction error was

minimized when superscores were used, as

compared to the other scoring methods. If

superscores reflected positive measurement

error—that is, an overestimate of one’s true

achievement level—then superscores would

predict students to earn higher grades in college

than what they actually earned, and this

overprediction would increase as the number of

retests increases. However, the results of the

study suggested exactly the opposite.

Why is this the case? One hypothesis is that

superscores and number of retests reflect not only

academic preparation but also a motivational

component. Specifically, students who are willing

to forgo multiple Saturdays to sit for a multiple-

hour test with the hope of maybe increasing their

score are also the students who are likely to ask

questions in their college courses, visit their

professor during office hours, and take advantage

of any extra credit opportunities to ensure the best

possible grade. Future research should explore

these hypotheses.

Another contribution of this study is the

evaluation of the diversity implications of

employing one scoring method versus another.

Interestingly, despite the fact that underserved

students are less likely to retest (Harmston &

Crouse, 2016), the superscoring method did not

result in a less diverse admitted class as

compared to the other three methods. In fact,

the gender, racial, and parental income

distributions of a simulated admitted class were

identical across the four scoring methods.

Current Study

The focus of the current study is to extend the

previous research with an emphasis on further

exploring the diversity implications of

superscoring. As mentioned above, underserved

students are less likely to retest as compared to

their affluent peers. For students who test only

once, superscoring has no impact on their ACT

Composite score. Only students who retest have

the potential to increase their ACT Composite

score through superscoring, and the magnitude of

this difference should be related to the number of

times the student retests, in general. With that in

mind, one potential concern or unintended

consequence of superscoring is that subgroup

differences will be exacerbated under this scoring

policy. The focus of the current study is to

investigate the extent to which superscoring

increases, decreases, or has no impact on

subgroup differences.

Method

Using data on the 2018 ACT-tested graduating

class, we compared the average ACT

Composite score for various student subgroups

based on their most recent ACT Composite

score as well as a superscore ACT Composite

score. We estimated subgroup performance

differences in terms of both:

1. Mean differences or unstandardized

differences (USTD): the difference between

the mean value in two groups

2. Standardized differences (STD): the difference

between the mean value in two groups,

divided by the overall standard deviation.

We estimated performance differences by the

following student characteristics: race/ethnicity,

gender, household income, and parental

education. These characteristics were self-

reported by students at the time they registered

to take the ACT.

Results

Retesting Rates. Samples sizes and retesting

rates for the 2018 ACT-tested graduating high

school class are summarized in Table 1.

Results are presented for the overall sample and

by the student subgroups of interest. The total

group consisted of over 1.9 million students. Of

those students, 44% took the ACT more than

once. As previously documented, we find that

minority students and students from lower socio-

economic households are less likely to retest.

ACT Research & Policy | Technical Brief | July 2019

For example, the retest rates for African

American and Hispanic students were 43% and

34%, respectively, as compared to 49% for both

White and Asian students. Differences by family

income and highest parental education level

were more pronounced. Students whose parents

did not attend college had a retest rate of 36%

compared to 62% for students whose highest

parental education level was more than a

bachelor’s degree.

Table 1. Proportion Retesting by Student Demographic Characteristics

Group N Proportion Retesting

Total

1,914,814 0.44

Gender

Male

893,609 0.41

Female

991,973 0.48

Missing

29,232 0.37

Race/ethnicity

African American

243,077 0.43

American Indian

15,449 0.36

White

996,712 0.49

Hispanic

307,358 0.34

Asian

91,899 0.49

Native Hawaiian/Pacific Islander

5,753 0.27

Multiracial

85,316 0.42

Missing

169,250 0.38

Annual family income

Less than $36,000 (Low)

353,315 0.40

$36,000 to $80,000 (Mid)

382,947 0.47

More than $80,000 (High)

498,300 0.60

Missing

680,252 0.34

Parental education level

No college

308,539 0.36

Some college

354,574 0.45

Bachelor's degree

418,863 0.55

Beyond bachelors

353,896 0.62

Missing

478,942 0.27

Note: The total number of students in the 2018 national ACT-tested cohort that are

reported here differs from that previously reported (e.g., ACT Condition of College and

Career Readiness Report and National Profile Report) due to a small number of

students since then being identified as being included more than once.

ACT Research & Policy | Technical Brief | July 2019

Table 2. Subgroup Unstandardized (USTD) and Standardized (STD) Differences in ACT Composite Scores by Scoring Method

Group

Most recent score Superscore Most recent score -

Superscore

M USTD STD M USTD STD ∆USTD ∆STD

Gender

Male 20.8 -0.1 -0.02 21.3 -0.2 -0.03 0.10 0.02

Missing 17.6 -3.3 -0.57 18.1 -3.4 -0.58 0.10 0.01

Female 20.9

21.5

Race/ethnicity

African American 16.9 -5.3 -0.91 17.4 -5.4 -0.92 0.10 0.00

American Indian 17.3 -4.9 -0.84 17.7 -5.1 -0.86 0.20 0.02

Hispanic 18.8 -3.4 -0.59 19.2 -3.6 -0.61 0.20 0.02

Asian* 24.5 2.3 0.40 25.1 2.3 0.39 0.00 -0.01

Native Hawaiian/Pacific Islander 18.2 -4.0 -0.69 18.5 -4.3 -0.73 0.30 0.04

Multiracial 21.1 -1.1 -0.19 21.6 -1.2 -0.20 0.10 0.01

Missing 19.8 -2.4 -0.41 20.2 -2.6 -0.44 0.20 0.03

White 22.2

22.8

Annual family income

Less than $36,000 (Low) 18.2 -5.7 -0.98 18.7 -5.9 -1.00 0.20 0.02

$36,000 to $80,000 (Mid) 20.7 -3.2 -0.55 21.3 -3.3 -0.56 0.10 0.01

Missing 20.0 -3.9 -0.67 20.4 -4.2 -0.71 0.30 0.04

More than $80,000 (High) 23.9

24.6

Parental education level

No college 17.9 -7.0 -1.21 18.3 -7.3 -1.24 0.30 0.03

Some college 19.8 -5.1 -0.88 20.4 -5.2 -0.88 0.10 0.00

Bachelor's degree 22.5 -2.4 -0.41 23.2 -2.4 -0.41 0.00 -0.01

Missing 18.9 -6.0 -1.03 19.2 -6.4 -1.08 0.40 0.05

Beyond bachelors 24.9 25.6

Note: ∆ refers to differences in unstandardized (USTD) and standardized (STD) subgroup differences for most recent score - superscore. Positive ∆ values

indicate that superscoring increases subgroup differences. Negative ∆ values indicate that superscoring reduces subgroup differences. Referent group is

italicized.

* For Asian students, the subgroup differences were positive; therefore, the signs for ∆USTD and ∆USD were reversed to maintain consistency in directionality.

ACT Research & Policy | Technical Brief | July 2019

Subgroup Differences. Table 2 provides the

means, USTDs, and STDs in ACT Composite

scores based on students’ most recent ACT

Composite score and their superscore ACT

Composite score for the student subgroups.

The difference in USTDs and STDs for the two

scoring methods—most recent versus

superscoring—was calculated to directly test

whether superscoring resulted in larger, smaller,

or similar subgroup differences as compared to

using students’ most recent score. Those results

are presented in the last two columns of Table 2.

The results indicate that superscoring increased

subgroup differences marginally. On average,

USTDs are 0.17 larger (on a 1 to 36 scale) for

superscores as compared to the most recent

scores (differences range from 0.00 to 0.40

among the subgroups examined; refer to ∆USTD

column).

For example, the average ACT Composite score

for African American students is 5.3 points lower

than White students (16.9 versus 22.2) when

based on the most recent ACT Composite

score. Comparatively, the average ACT

Composite score for African American students

is 5.4 points lower than White students (17.4

versus 22.8) when based on a superscore ACT

Composite score, resulting in a difference in

USTD of 0.10 (5.4 minus 5.3).

In terms of STDs, superscoring increases

subgroup differences by 0.02, on average

(differences ranging from -.01 to .05; refer to

∆STD column), representing a very small effect.

The largest effects were found for students with

missing data for either household income or

parental education level. Building off the

previous example comparing African American

students’ performance to White students’

performance, the results indicate no change in

STD with the implementation of superscoring.

Subgroup Differences by Number of Testing

Occasions. The next set of analyses explored

subgroup differences by scoring method

controlling for the number of testing occasions.

The rationale for these additional analyses was

to be able to tease apart subgroup differences

from differential retesting rates, given that

undeserved students are less likely to retest.

Table 3 provides the distribution of students in

terms of the number of testing occasions along

with the average ACT Composite score based

on most recent score and superscoring. For this

sample, 56% of the sample took the ACT only

once. Among retesters (44%), the breakdown

by the number of testing occasions was as

follow: 24% tested twice, 11% tested three

times, and 9% tested four or more times.

Table 3. Mean ACT Composite Scores by Scoring Method and Number of Times Tested

Number of times tested N (%)

Most recent score

Superscore

Mean SD Mean SD

1 1,064,222 (56%) 19.3 5.5 19.3 5.5

2 465,650 (24%) 22.0 5.7 22.9 5.6

3 215,527 (11%) 23.3 5.5 24.7 5.3

4 or more 169,415 (9%) 23.9 5.1 25.6 4.8

Total 1,914,814 (100%) 20.8 5.8 21.3 5.9

Note: SD = standard deviation. The total number of students in the 2018 national ACT-tested cohort that are reported here

differs from that previously reported (e.g., ACT Condition of College and Career Readiness Report and National Profile

Report) due to a small number of students since then being identified as being included more than once.

ACT Research & Policy | Technical Brief | July 2019

Students who tested more often tended to have

higher ACT Composite scores. Moreover, the

difference between the average ACT Composite

score based on the most recent score as

compared to superscoring increased as the

number of testing occasions increased, as

expected. For example, for students who tested

once, the average ACT Composite score was

19.3. For students who tested twice, the

difference in the average Composite score

between the two methods was 0.9 (22.9 minus

22.0). For students who tested four or more

times, this difference increased to 1.7. The

results broken out by student subgroups are

provided in the Appendix.

Table 4. Scoring Method Differences in USTD and STD in ACT Composite Score by Number of Times Tested

Group

Number of Times Tested

One Two Three Four or more

∆USTD ∆STD ∆USTD ∆STD ∆USTD ∆STD ∆USTD ∆STD

Gender

Male*

0.00 0.00 0.00 0.00 0.10 0.02 0.10 0.02

Missing 0.00 0.00 -0.30 -0.04 -0.40 -0.06 -0.30 -0.05

Female

Race/ethnicity

African American 0.00 0.00 -0.10 0.00 -0.10 0.02 -0.30 0.01

American Indian 0.00 0.00 -0.10 0.00 -0.10 0.01 -0.20 0.00

Hispanic 0.00 0.00 0.00 0.01 0.00 0.02 -0.10 0.01

Asian* 0.00 0.00 0.00 0.01 0.00 0.01 0.10 0.04

Native Hawaiian/Pacific Islander

0.00 0.00 0.00 0.01 0.00 0.01 0.00 0.01

Multiracial 0.00 0.00 -0.10 -0.01 0.00 0.01 -0.10 -0.01

Missing 0.00 0.00 0.00 0.01 0.00 0.00 -0.10 -0.01

White

Annual family income

Less than $36,000 (Low) 0.00 0.00 -0.10 0.00 -0.10 0.02 -0.20 0.01

$36,000 to $80,000 (Mid) 0.00 0.00 -0.10 -0.01 -0.10 0.00 0.00 0.03

Missing 0.00 0.00 -0.20 -0.03 0.00 0.01 0.00 0.01

More than $80,000 (High)

Parental education level

No college 0.00 0.00 -0.10 0.00 -0.20 0.01 -0.10 0.04

Some college 0.00 0.00 -0.10 0.00 0.00 0.03 -0.10 0.03

Bachelor's degree 0.00 0.00 -0.10 -0.01 0.00 0.01 0.00 0.02

Missing 0.00 0.00 -0.10 0.00 0.00 0.02 -0.10 0.00

Beyond bachelors

Note: ∆ refers to differences in unstandardized (USTD) and standardized (STD) subgroup differences for most recent score -

superscore. Positive ∆ values indicate that superscoring increases subgroup differences. Negative ∆ values indicate that

superscoring reduces subgroup differences. Referent group is italicized.

* For Asian students and males, the subgroup differences were positive; therefore, the signs for ∆USTD and ∆STD were

reversed to maintain consistency in directionality.

ACT Research & Policy | Technical Brief | July 2019

Table 4 provides the difference (∆) in USTD and

STD in ACT Composite score for the two scoring

methods by the number of testing occasions.

The directionality of the subgroup differences

were the same as those suggested in Table 2,

except for gender where ACT scores were

higher on average for males than for females

among students who took the ACT two, three, or

four or more times. As discussed, the most

recent and superscore results are identical for

students who tested only once. When the results

are disaggregated by the number of testing

occasions, we see even smaller increases

attributed to superscoring than previously

described and often the results reverse where

superscoring results in smaller subgroup

differences (as indicated by negative values).

For students who tested twice, USTDs are 0.09

smaller on average (ranging from -.30 to .00)

and STDs are the same (0.00 on average;

differences ranging from -.04 to .01) when based

on superscoring as compared to when based on

the most recent score. For students who tested

three times, USTDs are 0.06 smaller (ranging

from -.40 to .01) and STDs are 0.01 higher, on

average (differences ranging from -.06 to .03).

For students who tested four or more times,

USTDs are 0.09 smaller (ranging from -.30 to

.01) and STDs are 0.01 higher, on average

(differences ranging from -.05 to .04).

Discussion

In sum, the results indicate that subgroup

differences are largely unaffected by the two

scoring policies examined in the current study—

most recent versus superscoring. Given that

students tend to improve their scores through

retesting and the high reliability of ACT scores, it

is not surprising that results based on a

student’s most recent test record are quite

similar to those based on superscoring. Also

contributing to the finding of small to no

differences based on superscoring is the

relatively low frequency (less than half of

students) of retesting overall and retesting more

than once.

The results also suggest that the slight increases

in USTDs and STDs can be attributed to

differences in retest rates among subgroups.

Analyses controlling for the number of times a

student retests indicated that subgroup differences

were more likely to decrease rather than increase

when superscoring was applied. These results

are very promising. If we improve retesting rates

among groups who are less likely to retest, such

as underserved students, through programs and

initiatives, these results suggest that superscoring

may help reduce subgroup differences.

For example, broader awareness that ACT

provides two fee waivers to low-income students

for national test administrations may help

promote retesting among underserved students.

However, awareness may not be sufficient.

ACT data indicate that low-income students who

higher no-show rates than students who pay to

take the ACT—22% for fee waiver students

versus 6% for students who pay for their

registration (Cruce, Hahn, & Metcalfe, 2017).

Future research should explore ways to not only

promote registering for the ACT via the fee

waiver program but also encourage fee waiver

students to show up for the test.

The results of the current study provide a preview

into the impact of superscoring on subgroups

differences. Superscoring had little to no effect on

subgroup differences and in some cases,

resulted in smaller subgroup differences when

the number of retests was held constant. Despite

these positive findings, the results may change if

retesting behavior changes significantly in the

future in terms of who retests and how often.

ACT Research & Policy | Technical Brief | July 2019

Notes

1. Research on the SAT found similar findings pertaining to superscoring (Boldt, Centra, & Courtney,

1986).

2. The reason why there is a slight increase in the USTD but no change in the STD is due to the fact

that the standard deviation of superscores is larger than the standard deviation of the most recent

scores (5.9 vs. 5.8, respectively).

References

ACT, Inc. (2009). ACT-SAT Concordance Tables. Iowa City, IA: ACT. Retrieved from the ACT website:

https://www.act.org/content/dam/act/unsecured/documents/ACTCollegeBoardJointStatement.pdf

ACT, Inc. (2017). 2017 Higher Education Research Digest. Iowa City, IA: ACT.

ACT, Inc. (unpublished manuscript). 2019 ACT’s Higher Education Score Use Survey. Iowa City, IA: ACT.

Boldt, R. F., Centra, J. A., & Courtney, R. G. (1986). The validity of various methods of treating multiple

SAT

scores (College Board Report No. RR-86-4). New York, NY: The College Board.

The College Board & ACT. (2018) Guide to the 2018 ACT/SAT Concordance. Retrieved from the ACT

website: https://www.act.org/content/dam/act/unsecured/documents/ACT-SAT-Concordance-

Information.pdf

Cruce, T. M., Hahn, R. W., & Metcalfe, R. D. (2017). Nudging to No Effect: Attempts to Improve the

College Entrance Exam Attendance Rates of Low-Income Students. Paper presented at the

annual meeting of the Association for Education Finance and Policy. Washington: D.C.

Harmston, M. & Crouse, J. (2016). Multiple testers: What do we know about them? Iowa City, IA: ACT.

Mattern, K., Radunzel, J., Bertling, M., & Ho, A. D. (2018). How should colleges treat multiple admissions

test scores? Educational Measurement: Issues and Practice, 37(3), 11–23.

Mattern, K., Radunzel, J., Bertling, M., & Ho, A. D. (2017). How should colleges treat multiple admissions

test scores? ACT Working Paper 2017-4. Iowa City, IA: ACT.

ACT Research & Policy | Technical Brief | July 2019

Appendix

Table A1. Mean ACT Composite Scores by Scoring Method, Number of Times Tested, and Student Subgroup

One Time Two Times Three Times Four or More Times

Group % Recent Superscore % Recent Superscore % Recent Superscore % Recent Superscore

All students 56% 19.3 19.3 24% 22.0 22.9 11% 23.3 24.7 9% 23.9 25.6

Gender

Male 59% 19.2 19.2 23% 22.3 23.2 10% 23.6 25.0 8% 24.1 25.9

Female 52% 19.4 19.4 26% 21.9 22.8 12% 23.2 24.5 10% 23.8 25.5

Missing 63% 16.5 16.5 25% 18.4 19.6 6% 21.1 22.8 5% 22.7 24.7

Race/ethnicity

African American 57% 16.1 16.1 25% 17.6 18.6 11% 18.3 19.7 8% 19.1 21.0

American Indian 64% 16.2 16.2 21% 18.1 19.1 8% 20.1 21.5 7% 21.6 23.4

White 51% 20.6 20.6 25% 23.3 24.2 13% 24.3 25.6 11% 24.7 26.3

Hispanic 66% 17.9 17.9 23% 20.2 21.1 7% 21.8 23.1 4% 22.5 24.2

Asian 51% 23.2 23.2 26% 25.5 26.4 13% 26.3 27.6 10% 26.4 28.1

Native

Hawaiian/Pacific

Islander

73% 17.0 17.0 18% 20.3 21.2 6% 22.7 24.0 3% 23.5 25.1

Multiracial 58% 19.8 19.8 24% 22.3 23.3 11% 23.5 24.8 7% 23.9 25.6

Missing 62% 18.1 18.1 22% 21.7 22.6 9% 23.6 24.9 7% 24.1 25.8

Annual family

income

Less than $36,000 60% 17.4 17.4 25% 19.0 19.9 9% 19.7 21.1 6% 20.5 22.4

$36,000 to

$80,000

53% 19.7 19.7 26% 21.4 22.3 12% 22.2 23.6 10% 22.9 24.6

More than

$80,000

40% 22.6 22.6 28% 24.5 25.3 17% 25.0 26.3 15% 25.1 26.8

Missing

66% 18.5 18.5 20% 21.8 22.8 8% 23.9 25.2 5% 24.6 26.3

Parental education

level

No college 64% 17.2 17.2 23% 18.7 19.7 8% 19.5 21.0 5% 20.6 22.4

Some college

55% 19.0 19.0 26% 20.4 21.4 11% 21.2 22.5 8% 21.9 23.7

Bachelor's degree 45% 21.3 21.3 28% 23.0 24.0 15% 23.8 25.1 13% 24.1 25.8

Beyond bachelor's 38% 23.7 23.7 29% 25.5 26.4 17% 25.9 27.2 15% 25.7 27.4

Missing 73% 17.9 17.9 18% 20.7 21.7 6% 23.0 24.3 4% 23.8 25.6

ACT Research & Policy | Technical Brief | July 2019

Acknowledgement

The authors thank Wayne Camara and Jeff Allen for their feedback on earlier versions of this brief.

Krista Mattern, PhD

Krista Mattern is a senior director in Validity and Efficacy Research whose research focuses on predicting

education and workplace success through evaluating the validity and fairness of cognitive and non-cognitive

measures. Also known for work in evaluating the efficacy of learning products to help improve intended learner

outcomes.

Justine Radunzel, PhD

Justine Radunzel is a principal research scientist in Validity and Efficacy Research specializing in postsecondary

outcomes research and validity evidence for the ACT test.