Of the markets for questions from the political compass test (https://manifold.markets/browse?topic=politcal-compass-test), which will have the lowest variance, such that Manifold agrees on it most?
Strongly disagree = 0
Disagree = 1
Agree = 2
Strongly agree = 3
Minimum variance is achieved if 100% of people vote for the same option. Ties will be broken in favor of the market with more participants.
Question closes when the markets in question close. Initial choices based on markets that currently have low variance.
Inspired by https://manifold.markets/GustavoMafra/which-political-compass-question-wi
1,000
3.00I looked at every single market with the label, and did the variance calculations for anything even remotely close (using the python3 statistics module). Here are the variance values for every question listed in this market's options as well as some other questions from the data set that were close:
"possessing marijuana for personal use should not be a criminal offence": Manifold says "strongly agree" with variance 0.614
"mothers may have careers but their first duty is to be homemakers": Manifold says "strongly disagree with variance 0.483
"you cannot be moral without being religious": Manifold says "strongly disagree" with variance 0.368
"Astrology accurately explains many things": Manifold says "strongly disagree" with variance 0.355
"it is important that my child's school instills religious values": Manifold says "strongly disagree" with variance 0.292
"no one can feel naturally homosexual": Manifold says "strongly disagree" with variance 0.198
"I'd always support my country, whether it was right or wrong.": Manifold says "strongly disagree" with variance 0.146
The winner, and the political compass question Manifold has the least variance and thus strongest agreement on: "strongly disagree" to "I'd always support my country, whether it was right or wrong."
@josh I must admit I underestimated the majority American user base of Manifold. For what it's worth, I feared more over-zealous patriotism.. But less variance than astrology! I stand corrected.
@GazDownright I'm not sure "what is Manifold's spot on the political compass test" constitutes research. the impact is probably minimal as long as discovery of the polls isn't primarily through this question
@GazDownright Yes, I did. It probably would have done so less if it started out with an option for every one of the questions, but I wasn't going to manually add options for every question including the ones that were incredibly unlikely to win.
That said, I'd hypothesize that the manner in which it's likely to skew those polls is by adding more people to them, more so than it'll skew the distribution of the answers. (It will do both, but more the former than the latter.)