Have etable put stars in the same cell of the coefficient

[Giuseppe Polito]

I use the following to generate the regression tables:

quietly etable, estimates(h_1_1 h_1_2 h_1_3 h_1_4 h_1_5 h_1_6 h_1_7 h_1_8) ///
cstat(_r_b) cstat(_r_se, nformat(%7.2f)) ///
mstat(N) mstat(r2_p, nformat(%7.4f)) mstat(chi2) stars(.1 "*" .05 "**" .01 "***", attach(_r_b)) ///
showstars showstarsnote column(estimates) title("Models for hypothesis H1")

Then

quietly collect export "$outdir/tables/h1.xlsx", replace

But in the resulting Excel file, coefficients and stars are in two separate columns.

I need them in the same column, because Word won't format them as I want to if they are in two different columns.

[Mara Sheftel]

Hi, I am facing a similar issue and wondered if you figured this out back a few months ago! Thanks

[Seungmin Lee]

Dear all, I have the same issue; significance stars are created in separate columns.

When I generate a regression table with "etable" command, it seems the default layout is as below.

collect layout (coleq#colname#result[_r_b _r_se] result[N r2]) (cmdset#stars) (), name(ETable)

I assume "cmdset#stars" option creates separate columns, so I tried the following layout instead. But now it generates a separate rows.

collect layout (coleq#colname#result[_r_b _r_se]#stars result[N r2]) (cmdset) (), name(ETable)

Any way to attach significant stars in a same column?

Thank you.

[Jeff Pitblado (StataCorp)]

Showing your code/example without the data to go with it makes it difficult for us to make suggestions and show that our ideas actually work properly.

collect stars dynamically labels results given a set of significance rules for the labels. The labels are created depending on stars style type: result or dimension. result creates new items for the stars labels and tags them with result[stars]. dimension creates new items for the stars labels and tags them with stars[label] and all other already-existing items are tagged with stars[value]. In general collect defaults to type result, but etable defaults to dimension -- this is to accommodate vertically aligning the numeric items independently/separately from the stars labels.

Use command collect stars to change the type.

Here is a short example, using the publicly available auto data.

* data and model setup
sysuse auto
regress mpg turn
estimates store m1
regress mpg turn trunk
estimates store m2

* table of estimates, showing stars
etable, estimates(m1 m2) column(estimates) showstars

* show current stars specification; notice the type is dimension and the labels are attached to _r_b
collect query stars
* change the type to result
collect stars, result
* show that the only thing that changed was the type
collect query stars

* replay table to see that the stars are now attached to the coefficients
collect preview

Here is the resulting table.

-------------------------------
                          mpg  
-------------------------------
Turn circle (ft.)      -0.761**
                        (0.131)
Trunk space (cu. ft.)   -0.316*
                        (0.134)
Intercept              55.820**
                        (4.354)
Number of observations       74
-------------------------------

[Seungmin Lee]

Jeff Pitblado (StataCorp) Thank you so much once again! Next time I will come up with a data example. "Collect" command seems a bit complicated, but is really interesting that allows greater flexibility! Will take some time to go through it more thoroughly.