Could you make your figure a little larger, it’s hard to read on a desktop. It might also be easier for the reader if each of the five arguments had a one-word name to keep track of the gist of their actual content.
“As you can see, the winner in Phase 2 was Argument 9 by a nose. Argument 9 was also the winner by a nose in Phase 1, and thus the winner overall.”
I don’t think this is quite right. Arguments 5 and 12 are very much within the confidence interval for Argument 9. Eyeballing it I would guess we can only be about 60% confident that argument 9 would do better again if you repeated the experiment.
I would summarise the results as follow:
All five arguments substantially outperformed the control, on average increasing giving by around 45%.
We also had some evidence that Arguments 5, 9 and 12 all outperformed Arguments 3 and 14, perhaps having about 30% more impact.
According to the contest rules, the “winner” is just the argument with the highest mean donation, if it statistically beats the control. It didn’t have to statistically beat the other arguments, and as you note it did not do so in this case.
But many won’t interpret it that way and further clarification would have been good, yes.
Thanks for doing this research, nice work.
Could you make your figure a little larger, it’s hard to read on a desktop. It might also be easier for the reader if each of the five arguments had a one-word name to keep track of the gist of their actual content.
I don’t think this is quite right. Arguments 5 and 12 are very much within the confidence interval for Argument 9. Eyeballing it I would guess we can only be about 60% confident that argument 9 would do better again if you repeated the experiment.
I would summarise the results as follow:
All five arguments substantially outperformed the control, on average increasing giving by around 45%.
We also had some evidence that Arguments 5, 9 and 12 all outperformed Arguments 3 and 14, perhaps having about 30% more impact.
Eric Schwitzgebel responded as follows to a similar comment on his wall:
But many won’t interpret it that way and further clarification would have been good, yes.
Edit: Schwitzgebel’s post actually had another title: “Contest Winner! A Philosophical Argument That Effectively Convinces Research Participants to Donate to Charity”