Dear GNU PSPP developers,

I am writing regarding a bug in GNU PSPP 1.6.2 affecting the display of odds 
ratios, relative risks, and their 95% confidence intervals in the CROSSTABS 
procedure.

I understand that this issue has already been fixed in later PSPP releases. 
However, PSPP 1.6.2 is still distributed with Debian 12, meaning that users of 
this version may continue to receive incorrect statistical output.

I reproduced the issue on Debian 12 and successfully applied the correction to 
src/language/stats/crosstabs.c, replacing entries[i] with entries[j] in the 
risk estimate output.

I also updated the corresponding regression test in 
tests/language/stats/crosstabs.at, rebuilt the Debian package, and verified the 
corrected results using real statistical data.

I have prepared a minimal patch, pspp-1.6.2-crosstabs-fix.patch, containing 
only these two changes. The patch has also been verified against the original 
GNU PSPP 1.6.2 source archive.

My intention is not to report a new bug, but to ask whether the project would 
consider publishing a maintenance release of the 1.6.x branch, such as PSPP 
1.6.3, incorporating the existing correction.

This would allow users who need to remain on PSPP 1.6.x to obtain a corrected 
official release without upgrading to PSPP 2.x.

I have attached the patch for your consideration.

Thank you for your work on GNU PSPP.

Kind regards,



José Antonio Alguacil Ortiz
Contacto:
Teléfono: 658 538 640 
e_mail: [email protected]
--- a/src/language/stats/crosstabs.c
+++ b/src/language/stats/crosstabs.c
@@ -1965,7 +1965,7 @@
         {
           indexes[0] = j;
           pivot_table_put (risk, indexes, risk->n_dimensions,
-                           pivot_value_new_number (entries[i]));
+                           pivot_value_new_number (entries[j]));
         }
     }
   indexes[1] = pivot_category_create_leaf (
--- a/tests/language/stats/crosstabs.at
+++ b/tests/language/stats/crosstabs.at
@@ -1894,9 +1894,9 @@
 Table: Risk Estimate
 ,Value,95% Confidence Interval,
 ,,Lower,Upper
-Odds Ratio for factor (Placebo / Aspirin),2.25,2.25,2.25
-For cohort disease = Yes,1.08,1.08,1.08
-For cohort disease = No,.99,.99,.99
+Odds Ratio for factor (Placebo / Aspirin),2.25,1.09,4.64
+For cohort disease = Yes,2.00,1.08,3.72
+For cohort disease = No,.89,.80,.99
 N of Valid Cases,250.00,,
 ])
 AT_CLEANUP

Reply via email to