thomblake comments on 2011 Less Wrong Census / Survey - Less Wrong

77 Post author: Yvain 01 November 2011 06:28PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (694)

You are viewing a single comment's thread. Show more comments above.

Comment author: Yvain 03 November 2011 01:38:23PM *  14 points [-]

From your perspective, that makes sense. From my perspective - I don't intend to ever look at this data. I'm going to import it into SPSS, have it crunch numbers for me, and come out with some result like "Less Wrong users are 65% libertarian" or like "Men are more likely to be socialist than women."

If you put "other" - and this applies to any of the questions, not just this one - you're pretty much wasting your vote unless someone else is going to sift through the data and be interested that this particular anonymous line of the spreadsheet believes in strong environmental protection but an otherwise free market.

Looking at the answers, I really shouldn't have allowed write-ins for any questions - I was kind of surprised how many people can't settle on a specific gender, even though the aim of the question was more to figure out how many men versus women are on here than to judge how people feel about society (I considered saying "sex" instead, but that has its own pitfalls and wouldn't have let me get the transgender info as easily. I'll do it that way next time.)

I was particularly harsh on the politics question because I know how strong the temptation is. I think next survey I'll give every question an "other" check box, but it will literally just be a check box and there will be no room to write anything in.

Comment author: thomblake 03 November 2011 03:53:55PM 3 points [-]

Seriously, dude, coding. Surely someone would be willing to volunteer to code a couple hundred open-ends. It should take like 5 minutes if you're willing to use broad brushstrokes. And if most of the raw data is made public, the later sifting for interesting tidbits is crowdsourced.

Comment author: Yvain 03 November 2011 07:43:14PM 2 points [-]

Well, sure, you could do that. But if I decided to hand-code all of the political write-ins into standard political terms like "liberal", "conservative", "etc", then all I'd end up with is a list of people's political preferences in a few bins of standard political terms.

Which is exactly what I have now when I don't allow write-ins. This way is easier for me and allows people to choose their bin themselves rather than have me try to guess whether some complicated philosophy is more conservative than libertarian or vice versa.

Comment author: dlthomas 03 November 2011 07:56:40PM 6 points [-]

But does not allow for the creation of new bins, if we spot different clusters.

Comment author: thomblake 03 November 2011 08:55:30PM 5 points [-]

What dlthomas said. If 20% of your respondents wrote in "anarchist", then you have a new punch.