The media in this post is not displayed to visitors. To view it, please go to the original post.

New political compass just dropped!

britmonkey.com/2020s-political…

(Screenshot shows my result. Surprising no-one, I'm 92% woke to 8% chud.)

reshared this

in reply to Charlie Stross

The media in this post is not displayed to visitors. To view it, please go to the original post.

I'm slightly surprised to be told that I'm Zohran Mamdani, but I'm not unhappy about it. I'm not sure why that 4th access is Davos vs Dubai, though -- it feels like a distinction without a difference.
in reply to Charlie Stross

The media in this post is not displayed to visitors. To view it, please go to the original post.

Hmmm, 4% more Chud (?) otherwise pretty darned close to yours.

I guess that shouldn't be too surprising either, from my point of view seeing as I follow you daily. Read(ing) your books, etc.

My extra chud might be a few I don't know answers. Don't know enough to give a solid answer.

bitcoin = scam
AI = scam
Quantum Computing = future scam, at least the way they're going to implement it, i.e. get rich quick schemed.

in reply to Charlie Stross

The media in this post is not displayed to visitors. To view it, please go to the original post.

I have serious issues with this delightful quiz. For instance, the first item (screenshot):

  1. Girls --> Women?
  2. What fucking scoring system is going on with this?
  3. Stop looking at my answer

Second item (screenshot): This is two questions, clearly two separate dimensions. Smooshing them together will produce mostly garbage responses.

There are already weird things elsewhere, like I suspect some of the "don't know" answers are scored as a neutral/medium number (which is seriously bad practice in most cases), but I also suspect there's a lot of "wiggle room" when measuring things like this. The psychometrics might not be quite as nightmarish as they look.

Edit: I take it back. This is a shitshow, from a psychometric perspective. And it still probably does a basically adequate job of kind of classifying people (especially those on a far end of the spectrum).

It's really hard to take online fun-quizzes when your job for 20+ years has involved heavy doses of psychometrics and survey design. I blame myself.

This entry was edited (Friday, June 26, 2026, 11:24 PM)
in reply to Proto Himbo European

These Likert-scale ideological quizzes are always a psychometric shitshow. Here's my thorough panning of one a few decades ago:

astoundingteam.com/2010/06/04/…

So I decided to roll my own. I tried to cut down on package dealing and make each response item as clear, direct, and unambiguous as possible. Unfortunately, I am a disorganized slob and the source code is lost to the sands of time. The Wayback Machine did however capture the questions. This is what I came up with:

web.archive.org/web/2016080507…

I didn't include a scoring system, per se. My desire was to use clustering to identify "factions" within my "electorate" but I didn't yet know how to code that. I did play around with the correlation matrix, between questions:

web.archive.org/web/2015050202…

and between respondents (identified by whatever name they give for their ideology):

web.archive.org/web/2015050202…

I did make a little PCA map:

flickr.com/photos/61180684@N07…

in reply to Lorraine Lee

OMG this is the most delightful thing I've read in days! I hope I don't sound patronizing when I say that you don't seem to have a ton of psychometrics background and yet you did kick-ass, very reasonable psychometrics. This is awesome!

If you happen to have the data lying around I'd be thrilled to extend what you did. I'm not a super-expert; my approach would probably be to start where you did and move in another couple of directions, probably

  1. EFA (exploratory factor analysis; honestly usually gives nearly identical results to PCA)
  2. Maybe some kind of clustering, like latent class analysis or (abusing the Likert scales) latent profile analysis

That's about it, because you did almost exactly what I think most analysts would have done, already. This is super cool.

And if you feel like redoing the survey (honestly, if I find time I could put it on my limesurvey instance, since the questions are saved), that would be pretty awesome, too. This is just fun, and I think potentially informative and insight-producing.

You had me at "I rolled my own" :D

Edit: forgot to respond to your review of that other quiz because I was excited about the one you made. I pretty much agree, and now I have a new word for something I have issues with: "package dealing." You nicely identified a consistent problem: the premise of the question was taken as a given when it should have been a question or two itself. They were trying to get at a multidimensional thing with conditionals built in, with a simplistic approach. I've seen that done in ways that maybe aren't awful, but it's never great, and this one was pretty bad.

This entry was edited (Saturday, June 27, 2026, 6:04 PM)
in reply to Proto Himbo European

If I have a data backup it's on my old PC which is somewhere else. I'll check for it next time I'm there. I did find the source code, had to massage it quite a bit to get it working, sans capcha, so hopefully not too many spam responses. I've run it once, so as of this writing there should be one completed survey on the system.

astoundingteam.com/survey/

in reply to Charlie Stross

Starting with this question:
The primary cause of crime is poverty and deprivation.

The quiz kinda goes off the rails.

Crime here is left undefined, and I know they mean property crime and crimes regularly prosecuted by the local DA, but in reality, most crime is committed by rich people who buy politicians to avoid prosecution. (See graph showing that wage theft far exceeds burglaries.).

There really should be an option that is, "I disagree with the premise of the question."