Talk:Free statistical software/Draft

From Citizendium
Jump to navigation Jump to search
This article has a Citable Version.
Main Article
Discussion
Related Articles  [?]
Bibliography  [?]
External Links  [?]
Citable Version  [?]
 
To learn how to update the categories for this article, see here. To update categories, edit the metadata template.
 Definition This article briefly describes what free software is available for conducting statistical analysis of data. [d] [e]
Checklist and Archives
 Workgroup categories Mathematics and Computers [Please add or review categories]
 Talk Archive 1  English language variant American English

Just starting this article, feel free to jump in and help or give advice. Gene Shackman 04:55, 8 March 2009 (UTC)

add link to my review?

Would anyone be willing to add a couple of sentences to my review of free statistical software? http://gsociology.icaap.org/methods/soft.html This could be added into the main article section on reviews, perhaps saying something like what I wrote on my webpage "Basically, for correlation and simple regression, all gave the same results. Some also had the same results for forward stepwise." I only ask because I've searched the web and can't find any other review showing that the packages get the same results and so should be statistically equivalent. This seems like it would be useful information to have. Thanks Gene Shackman 03:45, 14 March 2009 (UTC)

call for comments

Hi Gene, I saw your note on the forum so just popped into check it out. Its an interesting article and certainly good information. I do not know anything about statistical software so cannot really comment on the topic at hand. Should there may be be a little more of a historical perspective? At present it reads like a "current" review that might become dated quite rapidly.

With regard to the following paragraph:

"Before using any statistical packages, it is generally a good idea to have a solid background in Statistics. Then the packages can be used to the best advantage, for example, to choose the most appropriate test, to make sure all the necessary assumptions are met, so that the appropriate conclusions can be drawn."

While this makes sense it is too general to be useful for a reader. Shouldn't there be specific examples of why some packages are better than others for any particular test? My first question would be "what makes these programs different?". Also, "why are some better than others for a particular test?"

Chris Day 14:48, 22 March 2009 (UTC)

Thanks for the comments. About "historical" - well, while some of the links may get to be outdated here and there, I think much of what is currently in the article is fairly stable. The packages mentioned are ones that have been around for a number of years. The reviews, articles and websites with listings and reviews are also in sources that are fairly stable, for example, journals, newsletters. Most of the tutorials, email lists and faq pages have also been around for a number of years. Should I add something about that, that most of these things have been around for some time?
about specific examples, and why some package may be better than others in some procedure... That certainly would be useful and I'm getting to that part, haven't written it yet.
and this leads to one question that I proposed above. As far as I know, there is only one source comparing the outcomes of the different programs, to show whether they are similarly accurate and what their output looks like, and the source is, well, mine. I wrote several reviews, here, http://gsociology.icaap.org/methods/soft.html. But this is at a site I own, so am I allowed to refer to it? There is the one other paper about the accuracy of easyreg and I do mention that, but mine is the only comparison that I could find, and I've searched. Could I refer to it, or would someone else be kind enough to add something about it to this article?
Thanks for the feedback and I'm certainly open to any other suggestions. Gene Shackman 15:12, 22 March 2009 (UTC)
With regard to historical, I was not thinking about the stability of the links or content but more about the history of these programs. Why were they developed? Who developed them? That sort of thing. Chris Day 15:45, 22 March 2009 (UTC)
I'll look into that, see if I can add something. Gene Shackman 17:14, 22 March 2009 (UTC)
Good start. Now, I may very well be downloading some software, but I know enough statistics to know what I don't know. I suspect the history and examples of applicability will go together — the fact that someone built the software presumably indicates that they saw a need. Especially when that need is perceived in an application area (e.g., epidemiology, agricultural yield), rather than "have statistical freeware for the statistics class", the example becomes more concrete.
I have no answer to the problem of both distrust and overconfidence in statistical methods in the general population. A friend of mine is constantly complaining how political polls can't show anything with a small sample, and he will also complain about biased sensational media polls. I'm not sure if his distrust is more due to not knowing anything about sampling, or about things that are not statistical per se but poll design (e.g., Guttman vs. Likert vs. Mokkert scaling, randomization of question order to avoid suggestion, etc.) Howard C. Berkowitz 18:16, 22 March 2009 (UTC)

what the devil is an NGO?

This is used all through the article -- Larry doesn't like abbreviations, especially ones that we (read "I") don't understand. Please change it, explain it, or remove it. Thanks! Hayford Peirce 15:28, 22 March 2009 (UTC)

I added the full name in the first paragraph - non governmental organization, like UNESCO. Gene Shackman 17:18, 22 March 2009 (UTC)
Thanks! I've never hoid it used before, but I suppose it's common enough. (Just did a little checking, it really does need a hyphen....) Hayford Peirce 17:32, 22 March 2009 (UTC)

Thanks for changing Non Govern.... to non-govern.... Gene Shackman 20:35, 22 March 2009 (UTC)

Sociology

Just out of interest, why is this in the Sociology Workgroup? Chris Day 15:42, 22 March 2009 (UTC)

I suppose it's in sociology because I'm a sociologist, and sociology uses a lot of stat analysis. But I'm certainly happy to see other work groups added, psychology, some science ones, anything anyone is interested in. I just can't remember how at the moment. Gene Shackman 17:18, 22 March 2009 (UTC)
All you have to do is go to the Metadata template page (http://en.citizendium.org/wiki?title=Template:Free_statistical_software/Metadata&action=edit) and add their names to the Categories list, of which there are three. Anyone can do it, although I suppose that this method can be open to dispute, which it *occasionally* has been (as when Martin Cohen started adding Philosophy to a lot of articles so that he could then be an [intrusive] editor), so it should be done with intelligence and discretion. Hayford Peirce 17:38, 22 March 2009 (UTC)

I'd say it is a bad precedent to add the sociology workgroup because it uses statistics. That is true for tens of our workgroups and we can only add three. One exception would be if this is a subject of sociology, possibly like the freeware movement could be looked at from a sociological perspective. But I don't think this article is going in that direction? Chris Day 18:26, 22 March 2009 (UTC)

Hmm, I'm on the sociology and psychology work groups, but I don't think I'm on others. Um, how would this work then? Do I need to get put on some work group that this article would be appropriate for? Math? Computers? How would I get added to those groups? Do I need to be on a work group to write an article in that group? Gene Shackman 18:53, 22 March 2009 (UTC)

No, you can write about what ever you like. The only thing that you can't do is approve it. Chris Day 18:56, 22 March 2009 (UTC)

Added the computer workgroup. Gene Shackman 19:00, 22 March 2009 (UTC)