Comments on: Statistical Software Popularity on Google Scholar

By: Bob Muenchen

Bob Muenchen — Sat, 14 Apr 2012 13:37:04 +0000

Nice catch Julio! Since we only plotted the top 6 packages, then dropped SAS and SPSS to plot the next 6, BMDP didn\'t show up. In 2011 it\'s last place with only 554 articles.

By: Bob Muenchen

Bob Muenchen — Sat, 14 Apr 2012 12:24:20 +0000

I don't think Google Scholar has a "near" function as some software does, but it uses the minus sign to exclude things. That would work fine in this case, but we learned that you have to be careful not to make Google Scholar queries too complex. The logic seems to fall apart eventually. You can test this by adding a very large number of "or" conditions. The values should always increase but they can eventually decrease. That may have only happened when there were also some "and" conditions, so the test may not be as easy as that. We were totally surprised though that the logic failed ever given the popularity of Google. In this case the number of additional hits was tiny though.

By: Laura O'Grady

Laura O'Grady — Sat, 14 Apr 2012 05:52:02 +0000

I'm not a programmer but I wonder if there is a what to write in something that can exclude, "Plant availabe soil water" etc. like: If PASW near 'Plant availabe soil water' then skip. Probably not worth the effort other than as an intellectual exercise as it has already been pointed out the likely use of PASW is limited. I've had research articles sent back from an editor because I didn't state the version number of SPSS I was using (in case a bug is revealed later, which could call my analysis into question).

By: Julio

Julio — Sat, 14 Apr 2012 05:37:05 +0000

I think there is an erratum in your bash script: it is BMDP instead of BDMP. BDMP stands for 2,6-dimethylphenyl or another chemicals.

By: flo

flo — Fri, 13 Apr 2012 09:11:03 +0000

amazing post. love the ggplot!

By: Bob Muenchen

Bob Muenchen — Thu, 12 Apr 2012 20:05:15 +0000

I should have replied to Laura O\'Grady but started a separate thread by accident, so here I am replying to my own post. We looked at PASW excluding SPSS and got a small number of messy hits. They included Plant Available Soil Water and Pluent Abdominal Segment Width. So we left both graphs (here and in the popularity article) unchanged. It was definitely worth a try though.

By: Bob Muenchen

Bob Muenchen — Thu, 12 Apr 2012 19:48:43 +0000

That market share plot mirrors Fig. 7b in r4stats.com/popularity in which SPSS and SAS are left out.

By: David Smith

David Smith — Thu, 12 Apr 2012 16:17:55 +0000

I love your "market share" chart - I prefer that as a way of representing the data to the overlaid time series chart. I wondered why SAS wasn't represented though, until I realised the "JMP" segment represents both SAS and JMP. I'm surprised at the early dominance of Systat though -- and do you know why it doesn't appear in the chart at r4stats.com/popularity? Anyway, thanks for providing this background information on how the data were collected.

By: Laura O'Grady

Laura O'Grady — Thu, 12 Apr 2012 14:08:58 +0000

Agreed. I don't know anyone that stopped calling it SPSS. I'm a researcher and have been trained to be "very thorough" or else your work will be questioned.

By: Kris

Kris — Thu, 12 Apr 2012 14:00:40 +0000

I remember when it was named PASW, still everyone, in psychological research at least, called it SPSS. I would image that any researcher using PASW would\'ve wrote it like PASW SPSS in their articles.