Title stata.com
Tables Builder — Tables Builder
Description Menu Remarks and examples
Description
After you have collected results using collect get, collect:, etable, dtable, or table, use
the Tables Builder to create tables of the results. In the Tables Builder, you can
• choose which results go in a table,
• lay out the rows and columns,
• change how row and column headers are arranged,
• change the look of anything, and
• export a table to Microsoft Word, Microsoft Excel, PDF, HTML, L
A
T
E
X, SMCL, Markdown,
and plain text.
Menu
Statistics > Summaries, tables, and tests > Tables and collections > Build and style table
Remarks and examples stata.com
Remarks are presented under the following headings:
Overview
Laying out a table
Laying out a multiway table
Modifying the layout
Laying out stacked dimensions
Placing multiple results in a cell
Multiple tables
Changing row and column headers
Text/labels
Layout
Appearance
Show/hide factor-variable base levels and empty cells
Show/hide omitted coefficients
Changing cell/results appearance
Numeric formats
Borders and horizontal or vertical rules
Bold, italics, text color, and such
Cell color
Margins
Justification/alignment
Adding significance stars
Adding a custom table title
Changing table title appearance
Adding table notes
Changing table note appearance
Exporting a table
1
2 Tables Builder — Tables Builder
Advanced tools
Position of intercept
Automatic dimension levels
Rename dimension levels
Remap tags
Overview
The Tables Builder works with an existing collection. You create a collection by collecting results
from commands using collect get, collect, etable, dtable, and table. The Tables Builder
lets you choose which results go in a table, lay out the rows and columns of a table, change how
row and column headers are arranged, change the look of anything in the table, and export a table
to Microsoft Word, Microsoft Excel, PDF, HTML, L
A
T
E
X, SMCL, Markdown, and plain text.
You launch the Tables Builder by selecting the menu item Statistics > Summaries, tables, and
tests > Tables and collections > Build and style table
You can see the major components of the Tables Builder in this diagram.
To use the Builder, you will need a basic understanding of what a collection is. In particular,
if seeing the words Dimensions and Levels above leaves you confused, read Tags, dimensions, and
levels in [TABLES] Intro 2.
If you are creating a table only to understand your data, you probably just need to lay out your
table. If you are creating a table for publication or to otherwise share with others, that is likely to be
Tables Builder — Tables Builder 3
an iterative process. It may take several steps to get the layout you want, then some steps to get the
headers exactly as you want them, and yet more steps to adjust numeric formats and text styles.
Laying out a table
The first step in creating a table is deciding what goes on the table’s rows and columns. We
cannot help you with that. The second step is getting those things onto the rows and columns, and
the Builder can help with that. The dimensions listed in the Dimensions list represent categories of
values pulled from results you collected. Each line in the list is a dimension description followed by
the dimension name in parentheses.
Choose which category you want to place on the rows of your table by clicking on one of the
dimensions in the list. If you want all the values associated (tagged) with that dimension, then just
click on the to the left of the Rows box. If you do that, all the levels of that category become
potential rows in the table. And you will see a term added to the Rows box. That term will be
identified with the name of the dimension.
If you want only a subset of the dimension’s levels, then select them from the Levels list before
hitting the control. As with the dimensions, each line in the list is a level description followed
by the level name in parentheses. You can click one level, then shift-click another to select a range.
You can control-click levels to select levels that are not neighbors in the list. If you select levels,
their names will become part of the description in the added term.
Do not worry too much about selecting the levels; you can modify the selected levels later.
We say “potential rows” because you must identify specific stored values using your selected
dimensions, or the Builder cannot create a table. For a detailed discussion of what is required to
properly specify a layout, see How collect layout processes tag specifications in [TABLES] Collection
principles.
If your table can be represented by a single dimension, you will see results in the Preview pane,
and you are done with the basic layout. This might be the case if you are using the Builder to pretty
up the results from a one-way tabulation command such as table rep78.
More likely, you are creating a two-way, or even multiway, table. To create a two-way table, you
will need to put a dimension into the Columns box. Again, click on the dimension, optionally select
some levels, and then click on the beside the Columns box. A term for that dimension will appear
in the Columns box.
If two dimensions are enough to lay out your table, you will see a table with the results you want
in the Preview pane. If two dimensions are not enough for your table, you will see one of two things
in the Preview pane.
1. You might see a message with some suggestions for other dimensions that may help lay out
your table. If so, try adding one of those dimensions to one of the terms in your Rows or
Columns box. (See Laying out a multiway table.)
2. You might see a table with results you are not interested in, and either the rows or columns
have labels that may confuse you but certainly do not interest you. What has happened?
This is really the same problem. You need a multiway table to present your results, and you
have only laid out a two-way table. Let’s tell you how to proceed and then tell you why it
happened.
First, think about the table you are trying to create. One, or more, of the dimensions you have
not yet put into the Rows or Columns box is required to identify the super-rows or super-columns of
a multiway table.
4 Tables Builder — Tables Builder
If you cannot think of any dimensions that you should add, clear your current layout by clicking on
the button in the lower left of the Builder. Repeat the process of selecting and adding dimensions
to your layout, but this time carefully select only the levels you are interested in. Now you will see
a message in the Preview pane, and you can go back and proceed from item 1 above.
So what happened to create that useless table? The commands that collect results collect everything
by default, including things you may not want on your table. You may have noticed this when selecting
dimensions for your table. You may have seen levels like c1 and c2 that do not interest you. Regardless,
they are in the collection, as are the values they tag. When the Builder laid out your table by searching
over all the levels in the dimensions you selected for rows and columns, it found results that exactly
matched some of the row/column tag combinations. That happened because you asked for a two-
way table when you wanted to ask for a three-way or higher multiway table. Once you add the
dimensions for a multiway table, the results found for the mistaken two-way table will no longer be
found—they only match the unwanted two-way table. For more details than you probably want about
what is required to properly specify a layout, see How collect layout processes tag specifications in
[TABLES] Collection principles.
Laying out a multiway table
A multiway table has super-rows, super-columns, or both to present results that require more than
two dimensions. Consider a cross-tabulation of frequencies by the categorical variables sex, region,
and agegrp. Those counts form a cube with dimensions for each of the categorical variables. We
cannot put three-way results directly on a two-way table. What we can do is create a three-way
table using super-columns or super-rows for one of the dimensions. Here is what that looks like with
super-rows for sex:
Tables Builder — Tables Builder 5
. use https://www.stata-press.com/data/r18/nhanes2l
(Second National Health and Nutrition Examination Survey)
. table (sex agegrp) (region)
Region
NE MW S W Total
Sex
Male
Age group
20--29 204 340 290 282 1,116
30--39 169 199 200 202 770
40--49 122 182 175 131 610
50--59 146 141 170 145 602
60--69 290 333 381 365 1,369
70+ 87 115 116 130 448
Total 1,018 1,310 1,332 1,255 4,915
Female
Age group
20--29 240 344 299 321 1,204
30--39 165 234 243 210 852
40--49 124 194 190 154 662
50--59 146 166 215 162 689
60--69 297 388 430 376 1,491
70+ 106 138 144 150 538
Total 1,078 1,464 1,521 1,373 5,436
Total
Age group
20--29 444 684 589 603 2,320
30--39 334 433 443 412 1,622
40--49 246 376 365 285 1,272
50--59 292 307 385 307 1,291
60--69 587 721 811 741 2,860
70+ 193 253 260 280 986
Total 2,096 2,774 2,853 2,628 10,351
There are several ways to lay out a three-way table in the Builder.
Method 1. The easiest way is to start by laying out a two-way table using just two of
your three required dimensions. You do that exactly as we did in Laying out a table. At
that point, you probably have a message in the Preview pane that you need to add more
dimensions. Or you might have a useless table that is not what you want. Either is to be
expected. Ignore that and move on.
Time to add your third dimension. Click on the in one of the two terms you just created.
If you want a table with super-rows, click on the on the term in the Rows box. If you
want a table with super-columns, click on the on the term in the Columns box. From
the resulting drop-down list, select Edit. You will be presented with a Modify term dialog
that looks something like
6 Tables Builder — Tables Builder
This dialog box looks and acts a lot like the Dimensions and Levels list from the main
Builder window. Note that there will be a check mark in the box beside the dimension you
previously selected for the term. That means the checked dimension is part of this term. If
you had selected specific levels when creating the term, those will have check marks beside
their lines in the Levels list of the dialog.
Now add your third dimension—the one you have not yet used—to this term. To do that,
just click on the box beside the third dimension in the dialog. If you only want some of the
levels on your table, you can click on the boxes beside the levels you want on the table.
Now click on OK. You will see your term updated in the Rows or Columns box. More
importantly, if the three dimensions you have selected do indeed identify the table, you will
see a table in the Preview pane.
The term you modified will now look something like
where colname and result are the names of the dimensions you selected for the term.
Your dimension names will probably not be colname and result unless you happen to
be creating a comparative regression table. They will be the names of the dimensions that
define your table.
The term might also look like
if you specifically selected the levels frequency and percent from the levels of dimension
result.
The # is actually part of the language for specifying table layouts when you use the command
collect layout instead of the Tables Builder. # instructs layout or the Builder to interact
the levels of the dimensions. If you think about how dimensions in the Rows and Columns
boxes are used, they too are interacted. In the case of the rows and columns of a two-way
table, we want all possible pairings of the levels from the row and column dimensions. Those
pairings form the cells in the table. When we created the interaction in our term, we were
requesting all possible triplets of the levels of the three dimensions. The third dimension is
organized as super-rows or super-columns.
Tables Builder — Tables Builder 7
Method 2. We can create a term with an interaction directly from the Dimensions list. We
start as we always do. Pick a dimension, and possibly select levels; then, use one of the
controls to put it into the Rows or Columns box. Then, click on that newly added term.
It will be highlighted. Then, select the third dimension from the Dimension list, and click
on the control for the Rows or Columns box where the term is highlighted. The new
dimension will be added to the highlighted term, and it will become an interaction term.
Finally, add one or more other dimensions to the empty Rows or Columns box.
Other methods. We did not have to perform all the steps in method 1 or method 2 in
exactly the order as written. And we can mix and match the methods from method 1 and
method 2. Put one term into the Rows box and then another into the Columns box; then
use method 2 to create interaction terms. We can delete terms and start over.
Moreover, we can create four-way tables, five-way tables, all the way up to silly-way tables.
We can use method 1 and method 2 to create as complex a multiway table as we like.
The important thing is to get our the interaction terms that specify our rows, super-rows,
super-super-rows, . . . ; and our columns, super-columns, . . . built so that we get the table
we want.
Modifying the layout
There are four main things we might want to change in a layout:
1. Make the rows columns and the columns rows.
2. Make super-rows rows and rows super-rows.
3. Make super-columns columns and columns super-columns.
4. Add or remove levels from a dimension.
1. Make the rows columns and the columns rows. Swapping rows and columns is both a
common desire and easy. You might accidentally put a dimension on the columns and
realize that 87 cities makes for a truly wide table. Drag the cities dimension to the Rows box,
where you always wanted it, and drag whatever dimension is in the Rows box to the Columns box.
Done.
This works with terms that have interactions too. If you drag an interaction term from the Rows
box into the Columns box, you are simply converting the whole super-row/row structure into a
super-column/column structure. And vice versa when dragging from the Columns box to the Rows
box.
2. Make super-rows rows and rows super-rows. This is a little more subtle, and you may have
already encountered it if you tried your own example when we discussed multiway tables above. We
showed a table in Laying out a multiway table. It had levels of the dimension sex as super-rows and
levels of the dimension agegrp as rows. That makes the comparison of age groups within sex easy.
What if you really wanted to compare females and males within age groups?
We need to swap the dimensions that are on the rows and super-rows. First, click on the on the
term in the Rows box, and select Edit to launch the Modify term dialog. You would
see that the sex and agegrp dimensions are checked and that sex appears above agegrp. Simply
click-and-hold on the agegrp dimension, and drag it above the sex dimension. Or click-and-hold
on the sex dimension, and drag it below agegrp. Click on OK. The dimensions on the rows and
super-rows have been swapped. The rows for females and males are now adjacent.
8 Tables Builder — Tables Builder
That is the general idea, and it also works for multiway terms where more than two dimensions
are interacted.
3. Make super-columns columns and columns super-columns. This is really just step 2.
Reread step 2, and substitute “column” everywhere you see “row”.
4. Add or remove levels from a dimension. You have rows you do not want, or columns that
you want are missing. You may have done this to yourself when creating terms by selecting too few
or too many levels for a dimension. Or you may have a collection from table that makes pretty
draconian assumptions about which result dimensions you want from any command() options.
If you have been reading along, you may have guessed the answer to this one. Click on the
of the term for which you want to add or remove levels, and select Edit to launch the Modify term
dialog. In the dialog, click on the dimension for which you want to change the list of levels. Then
check or uncheck levels until all the levels you want are checked, and only the levels you want are
checked.
Laying out stacked dimensions
There are cases where you do not want to interact a dimension with another dimension on the rows
or columns as we did in Laying out a multiway table. Instead, you want to stack some results below
some other results. (Or perhaps stack them to the right if the original results are on the columns.)
Consider a table comparing two regressions. Each regression appears in a column, so you al-
ready have the term in the Columns box. The coefficients and perhaps their standard
errors and confidence intervals appear on the rows. The values for these statistics require that you
also know the covariates. That means the Rows box will have an interaction term that looks like
.
Suppose you also want to place some of the model-level statistics below the coefficients and their
statistics. Say you want the R -square, the model F statistic, and the p-value of the model F statistic.
To add these statistics to the bottom of each regression, click on Result (result) in the Dimensions
list, and then control-click on R-square (r2), Model F test (F), and Model test p-value (p) in the
Levels list. Then, after making sure that nothing is selected in the Rows box, click on the beside
the Rows box. You will see a new term added to the Rows box, and the model-level statistics will
be added to the bottom of the table.
Placing multiple results in a cell
Rather than stacking results or placing them on adjacent columns, you may prefer to place multiple
results in a single cell. For example, suppose you have a table comparing regression results. You
currently have in the Columns box and in the Rows box, but
you would rather place the confidence intervals in the same cell as the coefficients. You can create
this layout by creating a composite result from the coefficients and confidence intervals. Click on
Manage composite results, and provide a name for this new composite result; for example, you might
name this result coefci. Then, for the Composite result elements, you would select Coefficient
( r b) and 95% CI ( r ci). You can then lay out your table by interacting colname with the
level coefci of dimension result.
Tables Builder — Tables Builder 9
Multiple tables
We have thus far completely ignored the Tables box. It has just one job. Instead of creating
super-rows or super-columns, any dimension placed in the Tables box creates multiple tables—one
for each level of the dimension. If you place or create an interaction term in the Tables box, a new
table will be produced for all combinations of the levels in the term.
What is more, you already know how to use it. Placing dimensions into the Tables box or creating
interaction terms in the Tables box works exactly as it works in the Rows and Columns boxes. You
can also drag terms into the Tables box from the Rows or Columns boxes, and vice versa.
Changing row and column headers
Row headers are the descriptions of the rows that you see along the left of a table. Column headers
are the descriptions of the columns that you see along the top of a table. These headers are the
reader’s guide to understanding your table. If the headers are good, your table will be easy to read
and understand. If the headers are bad, readers will be left scratching their heads.
Perhaps you are lucky, and your variable labels along with the default result labels and the default
header composition will be just what you want. Perhaps not. If not, there is plenty of control for the
headers in the Builder. You just have to know where to look.
Let’s start with the easiest part, the text actually written in the headers.
Text/labels
You can change the label for any dimension on your table. This is the label for the dimension
itself, not for the levels of the dimension. It is the levels that actually form the rows and columns of
your table. Sometimes, the dimension label is shown in the header; sometimes, it is not. See Layout
for a discussion.
To change one or more of your dimension labels, click on the button Edit dimension labels to
launch the dimension labels dialog. There you can select dimensions, see their current label, and
change that label.
You can also change the labels for levels in a dimension. These labels almost always appear on
the table because they identify the rows and columns. An exception is comparative regression tables
that sometimes identify standard errors by placing them in parentheses and confidence intervals by
placing them in brackets. Such tables often dispense with labeling the statistics, the levels, altogether.
Regardless, if you want to change labels for levels, click on the Edit level labels button to launch
a dialog for editing labels. In that dialog, you can choose the dimension with the levels you want
to change, choose the level you want to change, see the current label for the level, and change that
label.
Layout
There are two components to header layout—what things are shown and hidden, and how the
labels and names are composed to form the header.
Let’s start with what is shown and hidden. In Text/labels, we alluded to cases where dimension
labels themselves might or might not be shown. Consider a dimension created from a birth sex
variable. The two levels in that dimension would likely be labeled Female and Male. Those labels
are self-explanatory and do not need any additional labeling to be clear. We do not need, and probably
do not want, the label for the dimension itself. We would be perfectly happy with row headers that
look like this:
10 Tables Builder — Tables Builder
Female ...
Male ...
Now consider a variable from a Likert scale. Say we are talking about websites, and the label
on the dimension is “Easy to navigate”. The levels are labeled “Strongly agree”, “Agree”, “Neutral”,
“Disagree”, and “Strongly disagree”. Those level labels do not mean anything by themselves. We
need the dimension label—“Easy to navigate”. We now need row headers that look like this:
Easy to navigate
Strongly agree ...
Agree ...
Neutral ...
Disagree ...
Strongly disagree ...
Click on the Show/hide header content to launch a dialog that lets you control which labels or
names are shown. You can choose to show labels, show names, or show nothing (hide). You can make
this decision for dimensions, for all levels in a dimension, or for individual levels in a dimension.
You can even set the default behavior across all dimensions and levels.
Once your header content is set, you can change how row and column headers are composed or
constructed. You do that by clicking on the Compose row headers or Compose column headers
button.
That word “composed” encompasses a lot of choices. Let’s consider a few.
Do you want cross-tabulation row headers that look like
Female East
Female West
Female North
Female South
Male East
...
or like
Female
East
West
North
South
Male
East
...
Then choose either Split elements across columns or Stack elements in a single column on the
Compose row headers dialog.
Do you want your factor-variable interactions to look like
Sex # Region
Female # East
Female # West
Female # North
Female # South
Male # East
...
Tables Builder — Tables Builder 11
or like
Sex
Region
Female
East
West
North
South
Male
East
...
Then choose between Compose factor-variable elements in a single cell and Split factor-variable
elements into separate cells on the Compose row headers dialog.
Similar choices can be made about column headers on the Compose column headers dialog.
There are many other changes to row and column headers you can make from these dialog boxes.
We suggest you launch the dialogs and explore.
Appearance
You can change many things about the appearance of row and column headers, including borders
and horizontal or vertical rules, bolding, italics, font color, cell color, margins, and justification.
To change any of these properties for the row headers, click on the Cell appearance styles button.
In the resulting dialog, click on one of the check boxes for the dimension and level rows, and
then select Table cell type (cell type) as the dimension. If you want to change the appearance of
row headers, select row-header for the level. Alternately, if you want to change the appearance of
column headers, select column-header for the level. Then click on any of the dialogs tabs—Borders,
Diagonals, Fonts, Shading, Margins, Alignments, or Formats—and make any appearance changes
you want on that tab.
You can even target the appearance change to specific rows or columns. You do that by choosing
a second dimension on the dialog. Choose one of the dimensions that is on the rows or columns of
your table. Then, select the level of the dimension for the specific row or column header you want
to change. Finally, make the appearance changes you want on one of the tabs.
Show/hide factor-variable base levels and empty cells
Many applications of factor variables require that one of the levels in the variable be declared a
base level. In those applications, nothing is estimated for that level, making its coefficient effectively
0. By default, tables produced by the Builder include base levels for factor variables in the table with
0s for coefficients or means and blanks for standard errors and other statistics about the coefficient
or mean.
If you would rather have base categories dropped from a table, click on the Show/hide coefficient
styles button. On the resulting dialog, select Show base levels for factor variables. You then have
several choices for which base levels are shown. You can show base levels for factor variables but
not for interactions, show base levels for both factor variables and for interaction, or remove all base
levels.
Factor variables can be interacted, and sometimes there are no observations in one of those
interactions. Consider the interaction of race (which includes Aleut Eskimo) with city regions in Los
Angeles (LA). If there are no Aleutians residing in East LA, then we say that cell of the interaction
is empty.
12 Tables Builder — Tables Builder
As with base categories, you can control whether empty cells are shown or hidden on the dialog
launched by the Show/hide coefficient styles button.
We put this discussion in Changing row and column headers because showing or hiding factor-
variable base levels adds or removes entire rows or columns from the table, including their headers.
Show/hide omitted coefficients
Regressions and other estimators require that covariates not be collinear. If they are, this is flagged
in the output with a 0 coefficient and an (omitted) note. By default, tables produced by the Builder
include collinear covariates with 0s for coefficients or means, and blanks for standard errors and other
statistics about the coefficient or mean.
You can specify that collinear covariates instead be dropped from the results by clicking on the
Show/hide coefficient styles button. On the resulting dialog, select Show omitted coefficients, and
then click on the Off radio button.
Changing cell/results appearance
You can change just about anything about how the values in your table look—numeric formats,
borders, horizontal and vertical rules, bolding, italics, font, text color, cell color, margins, justification,
and more. This is all done in the dialog launched by the Cell appearance styles button.
If you want to change the default look of everything on the table, including the headers, click
on the radio button labeled Edit base style. Then click on one of the tabs—Borders, Diagonals,
Fonts, Shading, Margins, Alignments, or Formats. Whatever changes you make on those tabs will
apply to all text or cells throughout the entire table, both cells in the body of the table and in the
headers. Because you are modifying the base appearance, if you have previously made changes to
more specific tags (dimensions and levels), those changes will still be applied.
To make changes that override the default appearance for everything, first click on the radio button
labeled Edit styles for specified tags.
If you want to change the appearance of only the results and cells in the body of the table, and
not the headers, then select the dimension Table cell type (cell type) and select the level item. Now
any changes you make on the other tabs applies to all results and cells in the main body of the table,
and not to any of the headers.
You can be specific about which results are affected by the appearance changes you make on the
other tabs. For example, if your table has regression results, you could select the Result (result)
dimension, then select Coefficient ( r b). Any changes you make after that affect only the coefficients.
You might then click on the Formats tab and change the format type to Fixed numeric with 2 digits
to the right of the decimal. You could make the same changes to the Std. error ( r se) and 95%
CI ( r ci) by repeating the process on those levels.
You can get even more specific. So far, we have picked only one dimension. The Main tab of
the dialog allows you to pick up to 10 dimensions. Continuing with the regression example from
the prior paragraph, after selecting Coefficient ( r b) from the Result (result) dimension, we might
now pick a second dimension, say, Covariate names and column names (colname). Imagine we are
trying to highlight the results from one of the covariates in our regression. We would then select that
covariate from the Level drop-down list. So we have two things: the coefficient and the covariate
we want to emphasize. With those two tags chosen, any changes we make on the other tabs affect
only the coefficient and our chosen covariate. We can highlight those results any way we wish—bold,
italics, text color, cell color, etc. You could make the same changes to the Std. error ( r se) and
95% CI ( r ci) of the covariate by repeating the process on those levels.
Tables Builder — Tables Builder 13
It is important to realize that not all appearance edits can be rendered on all export formats. For
example, plain text (.txt) is called “plain” for a reason. Aside from numeric formats and some
horizontal and vertical rules, nothing from the appearance edits can be rendered in plain text. Most
appearance edits can be rendered in HTML, PDF, Microsoft Word, and Microsoft Excel. Many, but
fewer, can be rendered in L
A
T
E
X. No export format can render all the changes you can make on this
dialog.
Because the Preview pane renders one of the exports, you also will not be able to see all appearance
edits in the Preview pane. On Windows and Macintosh, the preview is HTML, so you will see most
appearance edits. On Linux, the preview is plain text, so you will see almost none of your edits.
Numeric formats
Make changes to numeric formats on the Formats tab. There you can also specify the delimiters
for confidence and credible intervals.
Borders and horizontal or vertical rules
Add or remove borders around cells on the Borders tab. You can create horizontal rules by
specifying borders only on the top or bottom of cells. You can create vertical rules by specifying
borders only on the left or right of cells.
Bold, italics, text color, and such
Make table text bold or italic, or change its color on the Fonts tab. You can also underline text
on the Fonts tab, or render it with a strikeout. You can even put text in all capitals or initial capitals.
There are also some special settings for L
A
T
E
X and SMCL export. You can specify the font family, but
that must be done in a way that is supported by your intended export format.
Cell color
Change background and foreground cell color on the Shading tab. You can also set fill patterns
on that tab.
Margins
Set margins for the cells on the Margins tab. You can add “in”, “cm”, or “pt” to any value you
type here to specify that the margin is to be in inches, centimeters, or printers points.
Justification/alignment
Align text to the left or right and to the top or bottom on the Alignments tab.
Adding significance stars
You can add significance stars to coefficients in the dialog launched by clicking on the Construct
significance stars. In fact, you can add any text you like to any result you like using rules on the
value of any result you like. You can even limit the application of the “stars” to selected tags. A tag
is defined by a dimension and one of its levels.
14 Tables Builder — Tables Builder
Adding a custom table title
You can add a custom table title in the dialog launched by clicking on the Custom table title.
Changing table title appearance
You can change the appearance of table titles in the dialog launched by the Table title styles
button. Click on the Fonts tab to change the font, size, text color, bolding, italics, and more. Click
on the Shading tab to change the background color, foreground color, and fill pattern.
Adding table notes
You can add table notes in the dialog launched by clicking on the Table notes.
Changing table note appearance
You can change the appearance of table notes in the dialog launched by the Table notes styles
button. Click on the Fonts tab to change the font, size, text color, bolding, italics, and more. Click
on the Shading tab to change the background color, foreground color, and fill pattern.
Exporting a table
Export your table by clicking on the Export ... button. You can choose from several export
formats—Microsoft Word, Microsoft Excel, PDF, HTML, L
A
T
E
X, SMCL, Markdown, and plain text.
Advanced tools
Position of intercept
When your table contains regression results, the intercept is placed by default after all the other
coefficients. If you prefer, you can have it placed before the other coefficients. Use the dialog launched
from the Intercept position button.
Automatic dimension levels
If you requested specific results when creating your collection, the meaning of a dimension used
alone in the Builder has been changed. For example, to collect results, you could type
. collect _r_b _r_ci: regress . . .
If you then add the Result (result) dimension to one of the Rows or Columns boxes without
choosing any levels from the Level list, that dimension will no longer represent the result dimension
and all of its levels. It will instead represent only the levels you specified on dimension result when
you collected the results.
Another way automatic levels are defined is when your collection was created by the table
command and you specified one or more command() options. table makes some choices on its own
about what you would like to see in your table.
Tables Builder — Tables Builder 15
Regardless of whether you requested specific results at collection time or table made some
choices for you, you can change these “automatic” dimension levels by clicking on the Automatic
dimension levels button. On that dialog, you can clear the automatic levels for any dimension. You
can redefine the automatic levels for any dimension. Or you can create a new list of automatic levels
for any dimension.
All of this can be tremendously helpful when typing commands to lay out a table. The Builder
makes it so easy to select levels while you are choosing dimensions for the rows or columns that
automatic levels are rarely helpful.
The most likely surprise you will encounter in the Builder occurs after a table command that
included regressions specified in the command() option. You might then place the Results (results)
dimension into the Rows or Columns boxes without selecting specific levels. If you do, only the
coefficients will be shown. That is because table set the automatic levels of dimension result to
be just the coefficients.
You could click on the Automatic dimension levels button and redefine the automatic levels for
result. But why? You can just click on the on the term, select Edit, and click on
whatever levels you want in the Modify term dialog.
Rename dimension levels
You can rename, or formally recode, any level from any dimension in the dialog launched by
clicking on the Recode dimension levels button. That would rarely be done from the Builder.
Remap tags
Another, even more advanced, operation you can perform from the Builder is to remap dimensions
and their tags. Click on the Remap tags button to launch a dialog. On that dialog, you can move
a level from one dimension to another dimension, create new dimensions and populate them from
existing levels in existing dimensions, and rename a dimension. You can even limit this remapping
to values where combinations of other tags are set.
Stata, Stata Press, and Mata are registered trademarks of StataCorp LLC. Stata and
Stata Press are registered trademarks with the World Intellectual Property Organization
of the United Nations. StataNow and NetCourseNow are trademarks of StataCorp
LLC. Other brand and product names are registered trademarks or trademarks of their
respective companies. Copyright
c
1985–2023 StataCorp LLC, College Station, TX,
USA. All rights reserved.
®
For suggested citations, see the FAQ on citing Stata documentation.
