All the Perl that's Practical to Extract and Report

Perl Survey Results with Pretty Pictures

schwern on 2007-10-13T15:12:58

[Edit] I've added Excel versions of the spreadsheets.

I've just finished giving a keynote at PPW showing the results of the Perl Survey. It's very preliminary data because 1) we just got the data a few days ago and 2) I failed statistics.

What we'd like is people who DO understand data visualization to play with the data and feed back their results to perlsurvey.org. On the results page there's a tarball for download containing the raw data dump as a CSV file and a MySQL dump. I also have a Postgres dump that contains some interesting views on the data to show, for example, who the Unix users are. In that directory there's also the two spreadsheets I showed off at the keynote containing some basic graphs of some of the data as well as a recording of the talk. I've provided the spreadsheets as both the original Numbers file and Excel. Numbers doesn't do ODF, but OpenOffice opens the Excel docs without mangling them too badly.

One of the things I'd like to emphasize is we don't know what slice of the actual Perl population the survey actually responded. We don't know the survey bias, we just know who responded. To give an idea, we took a quick poll of the people at PPW to see who in the room took the survey and it wasn't much.

We also took a number of informal polls of the folks at PPW. JCAP got pictures.

Not to keep harping on this ...

pudge on 2007-10-17T03:49:06

One of the things I'd like to emphasize is we don't know what slice of the actual Perl population the survey actually responded. We don't know the survey bias, we just know who responded.
Social science methodology tells us that this means your results have no meaning as applied to the "Perl community" as a whole. Or, as Wikipedia puts it, "Nonprobability sampling techniques cannot be used to infer from the sample to the general population."

That doesn't mean the study is completely useless, but it says nothing useful about the larger population, and since the goal is "to take a snapshot of the Perl world as it currently stands," the study doesn't seem to be capable of achieving that goal.