2
$\begingroup$

Possible Duplicate:
Should we allow more computing questions

I noticed there are a few discussions already about whether or not to allow for programming questions, like here and here. I assume most people currently on the side program either in R, Stata, SAS or SPSS. Yet, there is a multitude of other statistical software, and often researchers use the menus of any of those packages.

Personally, I would allow for questions regarding software, including questions about data manipulation. For R it has been argued that this belongs on SO, but for SAS, SPSS, Stata, there is no room there. Let alone for questions related to Minitab, Phylip, and other more specialized packages. GIS for example is often used in spatial statistics.

So not only questions like "where can I find this type of analysis in software Y?", but also questions like "I can't get my data read in, where am I going wrong?". I consider data manipulation an essential part of practical statistics, and would vote to allow those questions here as well.

On the other hand, this also holds the risk of diluting interesting theoretical discussions about statistics. Then again, most researchers looking for statistical advice aren't really helped with a theoretical rant. They often need some intuitive explanation of the tool, and a guide on how to start performing the analysis.

I add this question as I would like to extend the discussion (and decision) to all used software in statistics.

$\endgroup$
1
$\begingroup$

I'd like avoid actually answering your question here to briefly address your fourth paragraph, as I believe that's a paraphrase of my argument from this question. Please correct me if I'm mistaken.

I don't think that the danger is that we'll dilute discussions about statistical theory. There's a substantial middle ground between LaTeX-laden theoretical questions and "How do I read my data into SAS?" - questions where we actually talk about doing analysis. Not the deep-math theoretical aspects, not which tool to use or how to do X in Software Y, but actual analysis of data. It's that middle ground that I'm most concerned about preserving. I can read articles for theory, and I can RTFM for help with my software, but the only way I can get better at analysis is by getting feedback on my analyses and seeing what others are doing. As I work in relative isolation, that's often tough for me to do; here, I can get people from all over the world advising me on how to model my data. That's amazing. That's a hell of a lot more amazing than tech support or theoretical discussion. And that's what I want to preserve.

Sheesh, I feel like I'm posting political speeches today. Sorry.

$\endgroup$
  • $\begingroup$ I didn't paraphrase your argument, I would have mentioned it otherwise. But I agree, the line of thinking is about the same, and you definitely have a point there. $\endgroup$ – Joris Meys Sep 6 '10 at 16:10

Not the answer you're looking for? Browse other questions tagged .