​ Re: Multi response sets in PSPP

[Michał Dubrawski]

 

If developers are not interested in the task - we cannot complain about it. I have great respect for them, and appreciate it that they have spent so much of their time to make PSPP possible. I remember time when there was not GUI, or when the output was in plain text - the project has gone a long way thanks to the developers and all of the other people helping with the project (providing graphic, translations etc.).

​I believe that ​a few hundreds of dollars would be a good start but to hire someone to write CTables functionality for PSPP would be much higher cost. Without some official widespread crowdfunding campaign or/and support from some foundations it won't be possible.

Maybe promoting PSPP on Universities would attract some other developers for the project. It could for sure help with the development. Some time ago we have discussed here about the problem with Windows version. Maybe some Windows based developer would also be attracted? :-) But probably supporters of GNU movement do not use Windows that often.

In my opinion and according to my knowledge there is not a good GNU or open source solution for social scientists, students, non-government organizations or government to use for cross tabulation of survey results (especially with case weights). In R there are packages like survey (I personally use it and like it very much), but it is great to have a possibility to easily and nicely export results to spreadsheet like Gnumeric, Calc or Excel. You can then use Excel macros/formulas for example to test the proportions difference between groups with Z Test, the data is also easy accessible to make charts. I think PSPP with its cross tabulation is very close to meeting the minimum requirements of such solution. I sometimes export PSPP output to html file and open it with Excel. I know many people who have been trained to analyze data using SPSS during studies, and now they could not use their skills because their non-government organizations or small companies cannot afford buying SPSS.

My personal opinion is that because of its flexibility CTables command is much more complicated than typical user need. I think that much simpler version for example based on current cross tabulation command would be enough + possibility to view multi response sets (that alone would be great) + and maybe possibility to have more then one variable in the columns.

Going one step closer to the perfect solution (for me) I would add possibility to display summary numbers of total respondents in each column below the percentages. Perfect solution (for me) would be if we could not only have column percentages in columns but also have posibility for means, medians and standard deviation for each column (could be an alternative type of table). But that is dream solution for me, I would be happy to hear if this description would fit other users’ needs as well, or maybe you would need something totally different.

 

​warm regards,

Michal​