Update to -estout- available from SSC

[Ben Jann]

An update to the estout package is available from the SSC Archive. The update fixes a number of issues that were reported on GitHub. To install the update, type

. ssc install estout, replace

Ben

[River Huang]

Dear Ben, Many thanks for updating the useful estout package. I wonder if there is an equivalent option in estout can produce correlation matrix/table where different levels of significance is denoted by different numbers of stars? Please have a look at https://www.statalist.org/forums/for...commands/page6 (#82, and #83). Thanks!

[Ben Jann]

Do you mean something like this?

. sysuse nlsw88.dta, clear
. estpost corr age ttl_exp hours grade race wage, matrix
. esttab ., b(3) unstack nonum nomtitle not noobs compress

----------------------------------------------------------------------------------------
                 age      ttl_exp        hours        grade         race         wage   
----------------------------------------------------------------------------------------
age            1.000                                                                    
ttl_exp        0.124***     1.000                                                       
hours         -0.028        0.230***     1.000                                          
grade         -0.033        0.198***     0.087***     1.000                             
race          -0.059**      0.022        0.045*      -0.136***     1.000                
wage          -0.036        0.265***     0.159***     0.325***    -0.081***     1.000   
----------------------------------------------------------------------------------------
* p<0.05, ** p<0.01, *** p<0.001

[River Huang]

Dear Ben, Yes, it is exactly what I asked for. By the way, in cases where there are "a lot of" variables, it might be better the replace column names (say, age, ttl_exp, and so on) with numbers. A representative table is (from asdoc)

.

Is it possible for estout/esttab to have this option? Thanks.

[Ben Jann]

There is no automatic option to create such labels, but you can assign custom labels using options eqlabels() for the columns and coeflabels() for the rows. Here is an example that renames the columns and also includes the word "Variables" on the left:

. sysuse nlsw88.dta, clear
. estpost corr age ttl_exp hours grade race wage, matrix
. esttab ., b(3) unstack nonum nomtitle not noobs compress ///
>     eqlabels((1) (2) (3) (4) (5) (6), lhs("Variables"))

----------------------------------------------------------------------------------------
Variables        (1)          (2)          (3)          (4)          (5)          (6)   
----------------------------------------------------------------------------------------
age            1.000                                                                    
ttl_exp        0.124***     1.000                                                       
hours         -0.028        0.230***     1.000                                          
grade         -0.033        0.198***     0.087***     1.000                             
race          -0.059**      0.022        0.045*      -0.136***     1.000                
wage          -0.036        0.265***     0.159***     0.325***    -0.081***     1.000   
----------------------------------------------------------------------------------------
* p<0.05, ** p<0.01, *** p<0.001

To avoid excessive typing I would suggest putting together a small program that compiles the labels automatically. Example:

capture program drop mkcorrlbls
program define mkcorrlbls, rclass
    local vars: coleq e(b)
    local vars: list uniq vars
    local eqlabels
    local coeflabels
    local i 0
    foreach v of local vars {
        local ++i
        local eqlabels `eqlabels' (`i')
        local coeflabels `coeflabels' `v' "(`i') `v'"
    }
    return local eqlabels `eqlabels'
    return local coeflabels `coeflabels'
end

estpost corr age ttl_exp hours grade race wage, matrix
mkcorrlbls
esttab ., b(3) unstack nonum nomtitle not noobs compress varwidth(12) ///
    eqlabels(`r(eqlabels)', lhs("Variables")) ///
    coeflabels(`r(coeflabels)')

------------------------------------------------------------------------------------------
Variables          (1)          (2)          (3)          (4)          (5)          (6)   
------------------------------------------------------------------------------------------
(1) age          1.000                                                                    
(2) ttl_exp      0.124***     1.000                                                       
(3) hours       -0.028        0.230***     1.000                                          
(4) grade       -0.033        0.198***     0.087***     1.000                             
(5) race        -0.059**      0.022        0.045*      -0.136***     1.000                
(6) wage        -0.036        0.265***     0.159***     0.325***    -0.081***     1.000   
------------------------------------------------------------------------------------------
* p<0.05, ** p<0.01, *** p<0.001

[River Huang]

Dear Ben, Many thanks for this wonderful solution.

[River Huang]

Dear Ben, I did the following.

sysuse auto, clear
global all "price mpg rep78 headroom trunk weight length turn foreign"
estpost tabstat $all, listwise statistics(count mean sd p5 p25 p50 p75 p95) columns(statistics)
esttab using "log\esttab-t1b.rtf", replace cells("count mean(fmt(3)) sd(fmt(3)) p5(fmt(3)) p25(fmt(3)) p50(fmt(3)) p75(fmt(3)) p95(fmt(3))") noobs compress

and got the results (in word)

My question is: how to make the numbers of this table aligned? Thanks.

[Ben Jann]

you could add option alignment(r) to right-align the numbers.

[River Huang]

Dear Ben, Thanks. This is helpful.

[Nampuna D. Gultom]

Dear Ben Jann,

Thank you so much for creating and updating the estout package. I am sure many Stata users such as myself are using this package on a regular basis by now.

I know that this is an older post/thread, but I'd like to give it a shot since I have searched for hours and hours on Statalist and other sources, but I am still unable to find a solution to my problem. I sincerely apologies if I violate some of the rules on the Statalist FAQ.

Related to the discussion above (and please by all means if River Huang could contribute as well, I would be very much grateful), I would like to ask whether there is a way for me to display my own selected vars on the correlation matrix columns.

For instance, let's just say I would like to estpost correlate 20 different variables:

estpost corr var1-var20, matrix
esttab ., b(3) unstack nonum nomtitle not noobs compress eqlabels ((1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20), lhs("Variables"))

I would like the output to look instead of something like this:

Variables | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | ... | (20)
Var1
Var2
Var3
Var4
Var5
.
.
.
Var20

It will look like this below:

First Sequence

Variables | (1) | (2) | (3) | (4) | (5) | | (6) | (7) | (8) | (9) | (10) |
Var1
Var2
Var3
Var4
Var5
.
.
.
Var20

Second Sequence

Variables | (11) | (12) | (13) | (14) | (15) | (16) | (17) | (18) | (19) | (20) |
Var11
Var12
Var13
Var14
Var15
Var16
Var17
Var18
Var19
Var20

Basically, I was wondering if I could manually adjust the column of the correlation results (whether it's 5, 8, or 10 vars, as long as they are in the varlist; and in any order I want). Are there any ways to make output tables like that from estpost/esttab framework? Any help will be very much appreciated.

Best,
Nampuna