Dear all,
In Stata 14, I'm using a loop to build a matrix containing summary statistics for several variables, which is then written to an Excel spreadsheet using putexcel. I'm trying to format each number to look the way I want it to once it is written to the Excel file. The problem is that whenever I have a number large enough to have a comma, only the digits prior to the comma are written to the Excel file.
In the code below, the line
is supposed to take the mean of the variable and format it to display correctly. The macro `treatmean' is then added to a matrix which is written to the Excel file.
If I type
right after the
command, the macro displays correctly. However, if I type
(no quotation marks) the macro displays with a space instead of the comma. This leads me to believe that if I could fill in a matrix cell with "`treatmean'" rather than `treatmean', the resulting number would display correctly. But I cannot do that because I get a type mismatch.
The code is below. It is heavily commented because I was sharing the code with someone else who is not very familiar with Stata.
In Stata 14, I'm using a loop to build a matrix containing summary statistics for several variables, which is then written to an Excel spreadsheet using putexcel. I'm trying to format each number to look the way I want it to once it is written to the Excel file. The problem is that whenever I have a number large enough to have a comma, only the digits prior to the comma are written to the Excel file.
In the code below, the line
Code:
local treatmean: display %9.2fc `r(mean)'
If I type
Code:
di "`treatmean'"
Code:
local treatmean: display %9.2fc `r(mean)'
Code:
di `treatmean'
The code is below. It is heavily commented because I was sharing the code with someone else who is not very familiar with Stata.
Code:
local variables gastoaliment_percap FCS goodFCS
tokenize `variables' /* takes the first element of `variables' and assigns local
macro `1' to it, assigns `2' to y, etc.*/
local j: list sizeof variables
di `j'
matrix stats = J(`j', 6, .) /* blank matrix with `j' rows and 6 columns */
forvalues i = 1/`j' {
local l: var lab ``i'' /* variable label will go in our results matrix */
*summary stats
su ``i'' if treat2014 == 1 /* this becomes "su `1'" the first time through the loop. `1' gets
replaced with x because of the "tokenize" command above */
local treatmean: display %9.2fc `r(mean)' /* display %9.2fc changes each number into a number
that has a maximum of 9 columns, has commas separating 1,000s, and two decimal
places */
local treatsd: display %9.2fc `r(sd)'
su ``i'' if treat2014 == 0
local controlmean: display %9.2fc `r(mean)'
local controlsd: display %9.2fc `r(sd)'
di `treatmean'
* difference in means
qui reg ``i'' treat2014, cluster(Comunidad2)
local diff: display %9.2fc _b[treat]
local pvalue: display %9.2fc ttail(`e(df_r)', abs(_b[treat]/_se[treat]))*2
/* the ttail command returns the probability that T > t.
We want the probability that T > abs(t),
that is, the pvalue for a two-tailed test. So we have to multiple the
p-value by 2. Also, Stata uses the number of groups - 1 as the degrees of
freedom when you use a cluster-robust covariance matrix */
* fill in the matrix
matrix stats[`i',1] = `treatmean' /* mean */
matrix stats[`i',2] = `treatsd' /* st. dev. */
matrix stats[`i',3] = `controlmean' /* mean */
matrix stats[`i',4] = `controlsd' /* st. dev. */
matrix stats[`i',5] = `diff' /* difference in means */
matrix stats[`i',6] = `pvalue' /* p-value goes below the mean difference */
local ii = `i' + 11 // 11 rows of summary stats already created in earlier file
putexcel A`ii' = ("``i''") B`ii' = ("`l'") I`ii' = (`e(N)') ///
using guatemala_MDE, sheet("summary stats") modify
}
putexcel C12= matrix(stats) using guatemala_MDE, ///
sheet("summary stats") modify
Comment