wilcoxon signed rank test z-score

[Jia Gao]

Hi all, I'm stuck in a wilcoxon signed rank test z-score problem,can some one tell me how exactly command "signrank" calculate p.value and z , cause I found out the give different result compare to r when there are lot of zeros in the testing data,

here is the data , you can try on R (wilcox.test(Export,Export_U, alternative = "two.sided",paired = TRUE))and STATA (signrank Export=Export_U ), the result will be different , even the p.value

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(Export Export_U)
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
  .02333072           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
  .03802535           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0  .012133468
          0           0
          0           0
  .01906245           0
   .2665769           0
          0           0
          0           0
          0           0
.0013075314           0
          0           0
          0           0
          0           0
          0           0
  .19016194  .026820315
          0           0
          0           0
          0           0
          0           0
          0           0
          0    .4625474
          0           0
          0           0
          0           0
          0           0
          0           0
          0    .5120914
          0           0
 .006225195           0
          0           0
          0           0
  .24785425   .03528966
          0           0
          0           0
  .03483107           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0   .11168677
          0           0
          0   .05568282
          0   .01983493
          0           0
          0    .3025498
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          0           0
          1           0
          0  .012808466
          0           0
          0     .406177
          0           0
          0           0
          0 .0046266057
          0           0
          0           0
          0           0
          0 .0042638998
          0           0
          0           0
end

[Clyde Schechter]

The method of calculation is given in the Stata's online [R] manual. Go to the chapter on the -signrank- command and then click on the Methods and formulas link.

[Jia Gao]

The method of calculation is given in the Stata's online [R] manual. Go to the chapter on the -signrank- command and then click on the Methods and formulas link.

thanks Clyde, that's exactly what I'm looking for, you are always so amazing.

[Marcos Almeida]

About differences in Stata's - signrank - and R (and also SPSS), there is a thread you may wish to take a look (http://www.statalist.org/forums/foru...-statistic-why),.

I hope the remarks in #6 (in the above-mentioned thread) will be also helpful to you.

Best,

Marcos