summary table with t-tests

[Stuart Morrison]

I want to produce a LaTeX table which has means by group and then t-tests of the differences. Like this:

	mean (exp_a)	mean exp_b	diff	p-value
All
MBC=0
MBC=1
LKZ=0
LKZ=1
MBC

This feels like something that should be easy but I'm not finding it so. In this post Chuck Huber (StataCorp) outlines how to do this elegantly for tests between variables using table (command) (result) but that doesn't work if I want to include an if statement to test within a given variable. In this post https://www.statalist.org/forums/for...-test-to-latex @Andrew Musau outlines a solution using expand and esttab. But, this will be slow given my dataset is large and I need to because of the large number of comparisons I need to formulate and the number of reported observations will be wrong, etc. Is there any way to do this? I would especially like to find a way of doing it without using packages from ssc as I am working on a server which doesn't allow them to be quickly installed.

I tried something ugly like this:

collect clear
local fortable "exp_a =r(mu_1) exp_b=r(mu_2) Diff=(r(mu_2)-r(mu_1)) pvalue=r(p): ttest exp_a=exp_b"
table (command) (result),  command(`fortable')
preserve
keep if mbc==0
table  (command) (result) ,  command(`fortable')   append
restore
preserve
keep if mbc==1
table  (command) (result) ,  command(`fortable')   append
restore

and so on for a large number of groups. But this doesn't work. Thanks as always for any bright ideas. Stu.

[Attaullah Shah]

This can be done quite easily with asdocx, that can export to LaTeX, Word, Excel, and HTML. And installation of asdocx should not be an issue because it is not installed on the server, rather it is installed on the machine you are using. See this example:

sysuse auto, clear
asdocx ttest price == turn, by(foreign) template(ttest1)

  
 
------------------------------------------------------------------------------
Variables              Cat1       Cat2          dif.     t-value     p-value
------------------------------------------------------------------------------
foreign                                                                            
    0              6072.423     41.442      6030.981      19.967       0.000
    1              6384.682     35.409      6349.273      16.259       0.000
-------------------------------------------------------------------------------

There are numerous other features of asdocx which might be of interest to you https://fintechprofessor.com/asdocx/

proof.

. ttest price == turn if foreign == 0

Paired t test
------------------------------------------------------------------------------
Variable |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
   price |     104    6072.423    302.2182    3082.033    5473.045    6671.802
    turn |     104    41.44231    .3871602    3.948275    40.67447    42.21015
---------+--------------------------------------------------------------------
    diff |     104    6030.981    302.0502     3080.32    5431.935    6630.026
------------------------------------------------------------------------------
     mean(diff) = mean(price - turn)                              t =  19.9668
 Ho: mean(diff) = 0                              degrees of freedom =      103

 Ha: mean(diff) < 0           Ha: mean(diff) != 0           Ha: mean(diff) > 0
 Pr(T < t) = 1.0000         Pr(|T| > |t|) = 0.0000          Pr(T > t) = 0.0000

. ttest price == turn if foreign == 1

Paired t test
------------------------------------------------------------------------------
Variable |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
   price |      44    6384.682    390.6454    2591.248     5596.87    7172.493
    turn |      44    35.40909    .2236498    1.483525    34.95806    35.86012
---------+--------------------------------------------------------------------
    diff |      44    6349.273    390.5099    2590.349    5561.735    7136.811
------------------------------------------------------------------------------
     mean(diff) = mean(price - turn)                              t =  16.2589
 Ho: mean(diff) = 0                              degrees of freedom =       43

 Ha: mean(diff) < 0           Ha: mean(diff) != 0           Ha: mean(diff) > 0
 Pr(T < t) = 1.0000         Pr(|T| > |t|) = 0.0000          Pr(T > t) = 0.0000

[Stuart Morrison]

Thanks. Unfortunately, getting a subscription for the server I connect to for asdocx isn't possible so I am going to need to find another way.

[Rich Goldstein]

as the FAQ advises, just telling us something "did not work" is not very helpful; you might want to take a look at the 4th (of 7) blogs on the new -table- command; see ttps://blog.stata.com/2021/08/24/customizable-tables-in-stata-17-part-4-table-of-statistical-tests/

[Andrew Musau]

You should provide a reproducible example. It should be possible to loop, storing the results in a matrix and appending matrices. Then output using estout (SSC). I don't have Stata 17 to suggest a table solution.

webuse fuel, clear
ttest mpg1=mpg2
mat A=  r(mu_1), r(mu_2), `=r(mu_1)- r(mu_2)', r(p)
mat colnames A= Mean_1 Mean_2 Diff. P-value
mat rownames A= Mileage
esttab mat(A), mlab(none) substitute(_ " ")

Res.:

. ttest mpg1=mpg2

Paired t test
------------------------------------------------------------------------------
Variable |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
    mpg1 |      12          21    .7881701    2.730301    19.26525    22.73475
    mpg2 |      12       22.75    .9384465    3.250874    20.68449    24.81551
---------+--------------------------------------------------------------------
    diff |      12       -1.75    .7797144     2.70101    -3.46614   -.0338602
------------------------------------------------------------------------------
     mean(diff) = mean(mpg1 - mpg2)                               t =  -2.2444
 Ho: mean(diff) = 0                              degrees of freedom =       11

 Ha: mean(diff) < 0           Ha: mean(diff) != 0           Ha: mean(diff) > 0
 Pr(T < t) = 0.0232         Pr(|T| > |t|) = 0.0463          Pr(T > t) = 0.9768

.
. mat A=  r(mu_1), r(mu_2), `=r(mu_1)- r(mu_2)', r(p)

.
. mat colnames A= Mean_1 Mean_2 Diff. P-value

.
. mat rownames A= Mileage

.
. esttab mat(A), mlab(none) substitute(_ " ")

----------------------------------------------------------------
                   Mean 1       Mean 2        Diff.      P-value
----------------------------------------------------------------
Mileage                21        22.75        -1.75     .0463416
----------------------------------------------------------------

.

[Stuart Morrison]

Andrew Musau Thank you!! Yes. I was able to loop over the list of comparisons and use esttab as you suggested. Rich Goldstein I apologise for not being more explicit in my original question that the reason the approach I tried (following indeed the blog post you linked to) doesn't work is that I get the following output. Searching in Stata for cmdset reveals nothing, and other posts on this forum just suggest to use collect clear which I have already done.

Your layout specification does not uniquely match any items. Dimension cmdset might help uniquely match items.
. restore

[Attaullah Shah]

Just for reference, asdocx works well with Stata on a machine or a server. The installation files are stored on the machine you are working on. They have nothing to do with the server.