The bet was resolved with 6 yes votes, and 4 no votes, which means a victory for Carl Shulman. I will be sending Carl $10, as per our initial agreement.
I should note that this provided the maximum possible evidence for Oliver’s hypothesis given that outcome, and that as a result I update in his direction (although less so because of the small sample).
(I haven’t run this by Carl yet, but this is my current plan for how to interpret the incoming data)
Since our response rates where somewhat lower than expected (mostly because we chose an account that was friends with only one person from our sample, and so messages probably ended up in people’s secondary Inbox), we decided to only send messages until we get 10 responses to (1), since we don’t want to spam a ton of people with a somewhat shady looking question (I think two people expressed concern about conducting a poll like this).
Since our stopping criteria is 10 people, we will also stop if we get more than 7 yes responses, or more than 3 no responses, before we reach 10 people.
[Posting to note I have agreed to bet against Oliver on his proposed terms above.]
Why do people keep betting against Carl Shulman???
No shame if you lose, so much glory if you win
I wasn’t super-confident, and so far it looks neck-and-neck (albeit on a smaller and noisier dataset than we had hoped for, 10 instead of 20).
Any outcome yet?
And the results are in!
The bet was resolved with 6 yes votes, and 4 no votes, which means a victory for Carl Shulman. I will be sending Carl $10, as per our initial agreement.
I should note that this provided the maximum possible evidence for Oliver’s hypothesis given that outcome, and that as a result I update in his direction (although less so because of the small sample).
We had 8⁄10 responses, and just sent out messages to another batch to get the last two responses. Should be resolved soon.
What will you do about people who don’t reply to your messages?
(I haven’t run this by Carl yet, but this is my current plan for how to interpret the incoming data)
Since our response rates where somewhat lower than expected (mostly because we chose an account that was friends with only one person from our sample, and so messages probably ended up in people’s secondary Inbox), we decided to only send messages until we get 10 responses to (1), since we don’t want to spam a ton of people with a somewhat shady looking question (I think two people expressed concern about conducting a poll like this).
Since our stopping criteria is 10 people, we will also stop if we get more than 7 yes responses, or more than 3 no responses, before we reach 10 people.
I agree to this.
I’m interpreting this as “go until you get 20 ‘yes’ responses to (1) and then compare their responses to (2)”.