Figure 5.3 appears to be total log revenue rather than average, conflicts with text; model uses mean(log(revenue)) rather than log(mean(revenue)) about bds HOT 5 OPEN

Soon! Chapters are with production now. I'd say summer if I'm optimistic.

from bds.

Soon! It's with production, so hopefully early 2022.

from bds.

@shane-kercheval many thanks for this. Apologies for the delayed reply, I'm revising the book for a new addition (will be pretty cool; we're making it into a much more readable full-service text) and I just got to this section for revision.

You are absolutely correct! I got the order of operations mixed up and the plot descriptions also. The fixed analysis script is:

`
library(data.table)
ebay <- as.data.table(ps)
ebay <- ebay[,list(ssm.turns.off=mean(1-search.stays.on),
revenue=mean(revenue)),
by=c("dma","treatment_period")]
setnames(ebay, "treatment_period", "post.treat")
ebay <- as.data.frame(ebay)
head(ebay)

run the DiD analysis

did <- glm(log(revenue) ~ ssm.turns.off*post.treat, data=ebay)
coef(did)

library(sandwich)
library(lmtest)
coeftest(did, vcov=vcovCL(did, cluster=ebay$dma))`

Results are exactly you describe. I've also attached updated draft section in case you are curious. Comments/errata are highly welcome :-)
draft.pdf

from bds.

@mataddy, maybe not the best place for this question but do you have an idea of when the 2nd edition could be released? Thanks!

from bds.

I'm at risk of being a bit obnoxious here on Github but was curious if there was an update on the 2nd edition?

from bds.