Well, that is odd! Of the 25 poll entries with timestamps before 2013-04-20T20:00, 18 gave the first answer and 7 gave the second. Of the 30 entries after that time, 10 gave the first answer and 20 gave the second.
This is what we’d expect to see if people who reply early were overwhelmingly more likely to give the first answer. It’s also what we would see if someone did not like the way the poll was going and decided to rig it.
The entries which give the respondents’ usernames (of which there are only six) do not exhibit this change.
This is what we’d expect to see if people who reply early were overwhelmingly more likely to give the first answer. It’s also what we would see if someone did not like the way the poll was going and decided to rig it.
There is also the third alternative of a great comment defending option two showing up (or having been up-voted enough) at the time you mentioned, to sway “public opinion” in its direction. It seems highly likely that people would read the most visible comments (and be persuaded by them) before voting.
Now, I don’t know which comment was the most visible at (or right after) 2013-04-20T20:00, but it looks like PhilipL’s and buybuydandavis’ comments are the most probably candidates given their current karma scores. They are also a defense of the second option (or at least closer to the second than to the first).
yeah, but isn’t that because people often downvote or upvote in order to bring things back toward the middle? I’ve definitely seem people comment to that effect. Whereas polling shouldn’t have the same effect.
Take a look at the data. There’s a really clear inflection point. It’d be nice if the poll code gave us hashed IP addresses, /24s, or some other suitably privacy-protected way of checking against the more trivial sorts of poll-rigging, but it doesn’t.
Well, that is odd! Of the 25 poll entries with timestamps before 2013-04-20T20:00, 18 gave the first answer and 7 gave the second. Of the 30 entries after that time, 10 gave the first answer and 20 gave the second.
This is what we’d expect to see if people who reply early were overwhelmingly more likely to give the first answer. It’s also what we would see if someone did not like the way the poll was going and decided to rig it.
The entries which give the respondents’ usernames (of which there are only six) do not exhibit this change.
There is also the third alternative of a great comment defending option two showing up (or having been up-voted enough) at the time you mentioned, to sway “public opinion” in its direction. It seems highly likely that people would read the most visible comments (and be persuaded by them) before voting.
Now, I don’t know which comment was the most visible at (or right after) 2013-04-20T20:00, but it looks like PhilipL’s and buybuydandavis’ comments are the most probably candidates given their current karma scores. They are also a defense of the second option (or at least closer to the second than to the first).
seems like a really small sample size but yeah I’m surprised to see it end up exactly tied after how skewed it was at first.
Why? Comment voting goes like this all the time.
yeah, but isn’t that because people often downvote or upvote in order to bring things back toward the middle? I’ve definitely seem people comment to that effect. Whereas polling shouldn’t have the same effect.
Take a look at the data. There’s a really clear inflection point. It’d be nice if the poll code gave us hashed IP addresses, /24s, or some other suitably privacy-protected way of checking against the more trivial sorts of poll-rigging, but it doesn’t.