I am using Andy Field´s syntax for meta-analysis in SPSS. The syntax for basic meta-analysis works well, however the syntax for analysis of moderators (https://www.discoveringstatistics.com/repository/fieldgillett/Meta_Mod_r.sps) along with its launcher (https://www.discoveringstatistics.com/repository/fieldgillett/launch_meta_mod_r.sps) crashes with the error "subscript is out of range", giving number of "source line", which does not appear in the log (e.g. log ends with line 604, error message refers to line 642). The syntax always computes empty models, but asking (in the launcher) to perform moderator analysis, it crashes if I include other variable than one dummy variable (e. g. categorical with 3 levels, interval).
I found several discussions referring to this error in SPSS none with solution what might be wrong :-(.
Any ideas? If you got here, thanks for your time.
I have not inspected the code for this macro to understand whether it is limited as written to one dummy variate. Kindly allow me to point you to what I think is a much better-known and vetted add-on to SPSS (SAS as well), Andrew Hayes' PROCESS: http://www.processmacro.org/index.html
Good luck with your work!
I think that you might be better served using the metafor package in R - this is free, open sourced and can support a wider variety of models than those SPSS Macros. Viechtbauer's web pages provide a lot of support documentation also.
If I run self-made programs and I get that error anouncement, then it refers to the fact that I used an array that had been left undefined, or that has been defined with too few observations per variable, or that has been defined with a wrong number of dimensions. Arrays are (more-)dimensional variables, such as rs(ns, nri), meaning raw scores of subject number ns on variable nri. For instance, if I define rs(100, 10), then I get the error if I use more than 100 subjects, or more than 10 variables, or both. Inspect whether you defined somewhere in the SPSS-input too small a number of variables, too small a number of observations, or a wrong number of dimensions.
Dear Peter, dear David, thank you very much for your suggestions and advice. I got further on my own, but ran into different problems. I will try the other macro and I will post more about what worked and what did not for future reference. I appreciate your effort.
Dear Thom, I am quite familiar with SPSS, but statistics is not my everyday life, so I prefer not to learn another program... May be in future.
I would recommend RevMan, which is made available free by the Cochrane community. It's very comprehensive, dedicated to systematic reviews and meta analysis, and very easy to use. The manuals and tutorials are excellent and you can learn a lot about meta-analysis while using it.
http://community.cochrane.org/tools/review-production-tools/revman-5/revman-5-download
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS