pull down to refresh
30 sats \ 3 replies \ @Undisciplined 13 Oct \ on: Research in Public #04: Post fee influences post quality econ
I share your preference for the third specification, but it's nice to have the results from the fourth to gauge how large that effect is. To that end, people probably don't write the same kinds of posts in totally different territories, and even if they do there are different readers, so you might want to use pairwise user-territory fixed effects in place of the two separate fixed effects.
It's also a little confusing to focus on the results from specification 3, but then also state that the results aren't due to user differences.
I doubt the territory founder issue will be significant, since so few territories have changed hands. I was curious though if you excluded my ~econ posts from before I was the owner: i.e. is that designation static or dynamic?
You also might want to highlight how large these R-squareds are and what that implies for people who aren't trained in this stuff. It's interesting that so much variation is explained by these few variables.
Also, also, I was struck by how fat the tails are on the zap and comment distributions.
Again, this is a super cool project and I think you already have some very noteworthy results.
Again, this is a super cool project and I think you already have some very noteworthy results.
Imagine that, SN data being used for a future peer-reviewed paper~~ pinnacle of mainstream-ness.
reply
I doubt the territory founder issue will be significant, since so few territories have changed hands. I was curious though if you excluded my ~econ posts from before I was the owner: i.e. is that designation static or dynamic?
It's dynamic. So, the posts you made while you didn't own ~econ are included in the regression.
You also might want to highlight how large these R-squareds are and what that implies for people who aren't trained in this stuff. It's interesting that so much variation is explained by these few variables.
The high R squareds are mainly due to the fixed effects. Week/territory/user results in a pretty saturated model already.
It's also a little confusing to focus on the results from specification 3, but then also state that the results aren't due to user differences.
Thanks for the feedback. I'll definitely tighten up the writing for an actual manuscript, but for now I'm just trying to mind-dump the results as fast as I can haha.
Again, this is a super cool project and I think you already have some very noteworthy results.
Again, thanks. I'm actually very surprised at how easily the results have flowed out. Pretty much every specification I've tried has worked from the start. As I'm sure you know, that's not the case for most research projects.
reply
As I'm sure you know, that's not the case for most research projects.
Haha. Yeah, this actually happened on one of the most recent things I worked on and it was awesome.
reply