I’m thinking about A/B testing this by placing an Albertus Magnus portrait on my study desk and flipping a coin in the morning to determine if it will be visible for that day. If there is any effect, it should eventually show up in my Beeminder data (for studying logic, maths, and computer programming).
I predict that performing the experiment will result in improved productivity for the ‘control’ group, as compared to historical trends. I’m indecisive about the experimental days versus the control days.
Two weeks ago, I would have said 80%. I think that it is about four times as likely that you would report supporting results after two weeks, given that you were getting supporting results, than that you would report no experimental effect after only two weeks, given that the data did not support that prediction.
I’m thinking about A/B testing this by placing an Albertus Magnus portrait on my study desk and flipping a coin in the morning to determine if it will be visible for that day. If there is any effect, it should eventually show up in my Beeminder data (for studying logic, maths, and computer programming).
I would also love to see the results, can you message me when the experiment is complete?
Sure.
I predict that performing the experiment will result in improved productivity for the ‘control’ group, as compared to historical trends. I’m indecisive about the experimental days versus the control days.
With what probability?
Two weeks ago, I would have said 80%. I think that it is about four times as likely that you would report supporting results after two weeks, given that you were getting supporting results, than that you would report no experimental effect after only two weeks, given that the data did not support that prediction.
So now its 4:1 * 4:1 or 16:1, or about 95%
I would love to hear the results!
I’ll post the results in an open thread or something once I complete the experiment.