There are quite often requests on Statalist for ways to export correlation matrices as produced in Stata by correlate or pwcorr.
It is not always clear quite (1) what file format (2) what recipient software (3) what layout (4) what purpose is in mind, but many such requests seem close to wanting to display matrices in similar form in MS Word or MS Excel or possibly LaTeX or HTML. Rightly or wrongly, I often suspect some ritual display of matrices in Appendices that supervisors or reviewers insist on but almost no-one ever reads.
Be that as it may, many of the major tabulation or reporting commands will do a good job for you.
The point of this post is just to publicise a different way of exporting sets of correlations as Stata datasets in ways that might allow or even encourage further analyses, even just graphics.
corrci was written in the first instance to support confidence intervals for (Pearson) correlations using the atanh() transformation, often called Fisher's z transformation, but other bells and whistles were added at the time and later.
As is typical for such efforts, anyone interested in corrci is asked to look at the original paper but to use the latest version of the software (as I write, 2021).
Here is a token example, which extends to a quite common request to export several correlation matrices, typically one for each group of observations.
So, the idea is just to loop over each group and save each new set of results to a new dataset. Then loop over the datasets and put them together with append.
The result is trivial here, but the point is that more correlations and/or more groups wouldn't imply code that was much more complicated.
I'd particular commend graph dot as lending itself to easy and effective display (including comparison) of correlations.
To address one point before it is raised, corrci is not supportive of pairwise correlations in the sense of pwcorr, which I tend to distrust. It wouldn't be very difficult to clone corrci and edit the codeto support that, but such functionality is not on my to-do list.
It is not always clear quite (1) what file format (2) what recipient software (3) what layout (4) what purpose is in mind, but many such requests seem close to wanting to display matrices in similar form in MS Word or MS Excel or possibly LaTeX or HTML. Rightly or wrongly, I often suspect some ritual display of matrices in Appendices that supervisors or reviewers insist on but almost no-one ever reads.
Be that as it may, many of the major tabulation or reporting commands will do a good job for you.
The point of this post is just to publicise a different way of exporting sets of correlations as Stata datasets in ways that might allow or even encourage further analyses, even just graphics.
corrci was written in the first instance to support confidence intervals for (Pearson) correlations using the atanh() transformation, often called Fisher's z transformation, but other bells and whistles were added at the time and later.
As is typical for such efforts, anyone interested in corrci is asked to look at the original paper but to use the latest version of the software (as I write, 2021).
Code:
. search corrci, sj
Search of official help files, FAQs, Examples, and Stata Journals
SJ-21-3 pr0041_4 . . . . . . . . . . . . . . . . . Software update for corrci
(help corrci, corrcii if installed) . . . . . . . . . . . . N. J. Cox
Q3/21 SJ 21(3):847
improves explanation of the format() option and fixes a bug
concerning saving results to a new dataset
SJ-20-4 pr0041_3 . . . . . . . . . . . . . . . . . Software update for corrci
(help corrci, corrcii if installed) . . . . . . . . . . . . N. J. Cox
Q4/20 SJ 20(4):1028--1030
corrects code for a bias correction used if (and only if) the
fisher option is specified
SJ-17-3 pr0041_2 . . . . . . . . . . . . . . . . . Software update for corrci
(help corrci, corrcii if installed) . . . . . . . . . . . . N. J. Cox
Q3/17 SJ 17(3):779
new options added
SJ-10-4 pr0041_1 . . . . . . . . . . . . . . . . . Software update for corrci
(help corrci, corrcii if installed) . . . . . . . . . . . . N. J. Cox
Q4/10 SJ 10(4):691
update to fix corrci so that it always saves r-class results
SJ-8-3 pr0041 . Speaking Stata: Corr. with confidence, Fisher's z revisited
(help corrci, corrcii if installed) . . . . . . . . . . . . N. J. Cox
Q3/08 SJ 8(3):413--439
reviews Fisher's z transformation and its inverse, the
hyperbolic tangent, and reviews their use in inference
with correlations
Here is a token example, which extends to a quite common request to export several correlation matrices, typically one for each group of observations.
So, the idea is just to loop over each group and save each new set of results to a new dataset. Then loop over the datasets and put them together with append.
Code:
. sysuse auto, clear
(1978 automobile data)
. forval x = 0/1 {
2. corrci mpg weight displacement, saving(corr`x')
3. }
(obs=74)
correlations and 95% limits
mpg weight -0.807 -0.874 -0.710
mpg displacement -0.706 -0.804 -0.569
weight displacement 0.895 0.838 0.933
(obs=74)
correlations and 95% limits
mpg weight -0.807 -0.874 -0.710
mpg displacement -0.706 -0.804 -0.569
weight displacement 0.895 0.838 0.933
. use corr0 , clear
. gen foreign = 0
. l
+---------------------------------------------------------------------+
| var1 var2 r lower upper foreign |
|---------------------------------------------------------------------|
1. | mpg weight -.8071749 -.8744006 -.7095432 0 |
2. | mpg displacement -.7056426 -.8044354 -.5688671 0 |
3. | weight displacement .8948958 .8376901 .9326781 0 |
+---------------------------------------------------------------------+
. forval x = 1/1 {
2. append using corr`x'
3. replace foreign = `x' if missing(foreign)
4. }
(3 real changes made)
. l
+---------------------------------------------------------------------+
| var1 var2 r lower upper foreign |
|---------------------------------------------------------------------|
1. | mpg weight -.8071749 -.8744006 -.7095432 0 |
2. | mpg displacement -.7056426 -.8044354 -.5688671 0 |
3. | weight displacement .8948958 .8376901 .9326781 0 |
4. | mpg weight -.8071749 -.8744006 -.7095432 1 |
5. | mpg displacement -.7056426 -.8044354 -.5688671 1 |
|---------------------------------------------------------------------|
6. | weight displacement .8948958 .8376901 .9326781 1 |
+---------------------------------------------------------------------+
I'd particular commend graph dot as lending itself to easy and effective display (including comparison) of correlations.
To address one point before it is raised, corrci is not supportive of pairwise correlations in the sense of pwcorr, which I tend to distrust. It wouldn't be very difficult to clone corrci and edit the codeto support that, but such functionality is not on my to-do list.
Comment