I have 10 variables (var1-var10) for 10 days (day) for 10 individuals (id) in long format. However, I recreated an example with 3 variables, for 5 days for 3 individuals in dataex:
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input byte(id day) float(var1 var2 var3) 1 1 .7806329 .2003541 .9929246 1 2 .8852594 .7479552 .3049926 1 3 .573707 .22255337 .8362682 1 4 .2431343 .9553812 .1578373 1 5 .008077622 .8259419 .8112878 2 1 .4933739 .1188735 .36237845 2 2 .7138895 .55584866 .884728 2 3 .4031537 .3037445 .008933238 2 4 .1250264 .53452736 .7117536 2 5 .7928984 .3798434 .6441231 3 1 .25131217 .063004404 .9543635 3 2 .6858981 .6998435 .8498245 3 3 .9965313 .483217 .4657771 3 4 .9135284 .8579231 .13088849 3 5 .29056227 .25628337 .5243536 end
The best I can do is:
Code:
local varlist "var2 var3" foreach i in 1 2 3 { display _newline foreach a in `varlist' { qui pwcorr var1 `a' if id==`i' local `i'_`a'_r = r(rho) if r(rho) !=. & r(rho)!=0 { //in the actual dataset, var1 may not be available or may not vary. This is how I exclude them from the list. display "`i'_`a'_r:" ``i'_`a'_r' } } }
Code:
1_var2_r:-.54861605 1_var3_r:.04435157 2_var2_r:-.05894353 2_var3_r:.25521847 3_var2_r:.79714705 3_var3_r:-.61136201
Just to remind you that the example data is much smaller (3 x 5 x 3) than the actual data (10 x 10 x 10), so in the latter case it makes more sense to sort the correlation coefficients.
Thanks so much.
0 Response to Sort by correlation coefficient
Post a Comment