You just have to accept that the dudes are ruining your results literally no matter what you do if you donβt add a third option. Itβs forgone. Itβs too late. You must abandon hope permanently
Conversation
Itβs not a favor to your male followers (almost certainly >80% of ppl who will see it btw), itβs literally just the only thing you can do to salvage even a tiny amount of epistemic value from@the survey
5
37
women only, do you understand how public polls work on a site that's 80% male and 95% doesn't give a fuck about trading off their curiosity for your data quality?
- no17.9%
- stop mansplaining22.9%
- it's the men who r wrong26.4%
- abolish gender32.9%
140 votesFinal results
2
10
The question is which poll answer do you think the men are more likely to pick
- Y57.5%
- N42.5%
127 votesFinal results
3
6
they're likely to pick "not a woman/show results", along with all the women who are unsure or don't want to answer for any reason
1
9
See having experience in actual data collection Iβm aware that twitter polls are not fantastic epistemology to begin with and assumed all my male followers were aware enough of this not to come into my replies rending their clothes bc they canβt look at a poll for a day
3
19
getting mad at Curzi for pointing out that your poll is ruined isn't dissuading the hundred anonymous accounts actually ruining your polls
1
19
And no amount of poll construction will make a Twitter poll a truly good source of data
Itβs my poll, Iβm content with the results as they are, itβs all of you who are not, you deal with that on your own time and not in my mentions
2
11
There's pretty good methods I think, some twitter polls I trust more than some normie surveys I've done. You can measure the error rate by repeating polls, or honesty by doing variants on polls where you change wording or option styles.
1
6
I've done and tracked in a spreadsheet responses to over 3k polls, lots of them variations of quality tests, and at this point I feel pretty good at being able to estimate how reliable the answers to a poll are. It can be done!
1
5
But I definitely recommend including a results option, you'll get less warped data that way. You can check this by tweeting variants across time where some include results option and some don't, and other polls where u measure the relevant population difference.
i was aware of the limitations in tweeting this poll as i did and chose to do so anyway; i certainly did not expect to get this much feedback from men complaining about it
on the broad question of twitter polls i am skeptical about how well sampling limitations can be overcome
2
5
even the most broadly shared polls are not going to a particularly representative sample of the population bc twitter users are a particular brand; my polls are hazy and local and that's an acceptable known cost to me
1
2
Show replies