Oh, wow. Wow especially since in Germany, Facebook has been forced, through fines, to hire many moderators to respond to hate speech. If this finding holds, imagine the effect around in places Facebook has been terribly understaffed and unresponsive for years—Burma, Sri Lanka...https://twitter.com/elipariser/status/1031983483151892480 …
-
Show this thread
-
This Tweet is unavailable.
-
Replying to @ShuhBillSkee @zeynep
I am skeptical about the statistical inference here. Note that their robustness checks with alternative (more reasonable) choices of standard errors do not include the critical results based on Internet/FB outages. My uncharitable guess: it didn't remain significant.
2 replies 1 retweet 10 likes -
Replying to @deaneckles @ShuhBillSkee
Just the kind of topic internal Facebook data would be extraordinarily useful! Too many such studies are underpowered and/or not great causal design, but that's not really their fault. Hate speech/violence link is such an important Q but hard to study causally.
1 reply 0 retweets 11 likes -
This Tweet is unavailable.
-
Replying to @ShuhBillSkee @zeynep
Why would internal data be particularly useful here? Many threats to validity for this work are more basic than that.
1 reply 0 retweets 1 like -
Replying to @deaneckles @ShuhBillSkee
Linking the visibility/spread of hate speech to external events--dunno maybe using moderation levels/success as controls? Were there places where there was more/less effective moderation compared to other places? Anything else that constitutes a natural(ish) experiment?
1 reply 0 retweets 4 likes
Internal data doesn't solve all threats to validity (since there is almost certainly a feedback look, right—more organized anti-refugee groups, more Facebook postings?) but is probably the best shot at trying to distangle the many vectors.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.