Regression table

[robin ackermann]

In did different regressions and put them together with the "esttab" command. I also scaled them by standard deviation of an underlying variable. In the six created regression tables the Adj. R^2 and also the t-statistic value are calculated.

In my combined table created with the esttab command, I also want Stata to show me the R^2 and the t-statistic value in this table, but so far I did not come up with any solutions.

Here's my code:

reg bklevg inlevg
qui: estadd local YearFE "No" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg1

reghdfe bklevg inlevg logsales mktb prof tang, a(fyear) vce(robust)
qui: estadd local YearFE "Yes" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg2

reghdfe bklevg inlevg logsales mktb prof tang indmedbl divpayer, a(fyear) vce(robust)
qui: estadd local YearFE "Yes" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg3

reg mklevg inlevg
qui: estadd local YearFE "No" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg4

reghdfe mklevg inlevg logsales mktb prof tang, a(fyear) vce(robust)
qui: estadd local YearFE "Yes" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg5

reghdfe mklevg inlevg logsales mktb prof tang indmedbl divpayer, a(fyear) vce(robust)
qui: estadd local YearFE "Yes" , replace
qui: estadd local IndustryFE "No" , replace
eststo reg6

esttab reg1 reg2 reg3 reg4 reg5 reg6, transform (inlevg @*0.1926886 @ logsales @*2.339952 @ mktb @*2.4294071 @ prof @*0.33298033 @ tang @*0.23045626 @ indmedbl @*0.09862091 @ divpayer @*0.48950112 @)
label
cells(b(fmt(%4.2f)) se(par fmt(%3.2f))) starlevels(* 0.1 * 0.05 ** 0.01) ///
stats(YearFE IndustryFE N, fmt(%12.0fc %12.0fc %12.0fc) label("Year FE" "Industry FE" "Observations"))

[Andrew Musau]

estout is from SSC (FAQ Advice #12). In multiple regression, if you are doing inference with regard to a single parameter, you have the t-statistic. On the other hand, if you are doing inference with regard to multiple parameters, you have the F-statistic. In that regard, therefore, you will have multiple t-statistics, one corresponding to each parameter in the regression. Now, there is a direct link between coefficients, standard errors and t-statistics. If you give me a pair, I can tell you what the third statistic is. The correspondence is \(\frac{\text{Coefficient}}{\text{Standard error}}= \text{t-statistic}\). Now, since we always report coefficients, then you need to choose whether to report t-statistics or standard errors. In esttab, you specify these using the options -t- or -se-

esttab reg*, t

or

esttab reg*, se

For the R2 statistic, you include this within -scalars()- or -stats()-. See

ereturn list

to view how the statistics are stored as.

esttab reg*, t scalars(N r2)

More detailed discussions in:

help estout
help esttab