4 ms·

The full text is on Sci-hub: https://sci-hub.tw/10.1016/j.jaac.2019.04.020 https://sci-hub.tw/10.1016/j.jaac.2019.04.020 I'm extremely skeptical. I don't know

by ggreer 7y ago

The full text is on Sci-hub: https://sci-hub.tw/10.1016/j.jaac.2019.04.020 https://sci-hub.tw/10.1016/j.jaac.2019.04.020 I'm extremely skeptical. I don't know how many ways they carved up the data before they found a segment of the population with increased suicide rates at the right time. Also, the show is about a girl who kills herself, but the female rate of suicide didn't increase enough to be considered statistically significant. Lastly, the spike in suicides starts in March, which is a month before the show was released. Their claim only makes sense if you think that watching the trailer is enough to trigger suicides.

sigstoat 7y ago
> I don't know how many ways they carved up the data before they found a segment of the population with increased suicide rate at the right time. and they still reported 9 different categorizations (overall/male/female * 3 age groups), and could only find one with a change. it isn't clear to me if they corrected for the multiple comparisons they did report.
- ggreer 7y ago
  Right, but that doesn't tell me how many other ways they carved the data up and didn't report it in the paper. Maybe they initially chose different age groups (10-22, 23-50, >50) and found no effect. Then they tweaked the categories and tada, controversial paper!
  - sigstoat 7y ago
    > Right, but that doesn't tell me how many other ways they carved the data up and didn't report it in the paper. i understand that and i don't believe i contradicted you, rather i was emphasizing: they _admitted_ to 9 different carvings and only got 1 interesting one. what they did under the hood had to be even worse!
malandrew 7y ago
Maybe I missed it in the article, but did they check the Netflix viewing history of those that committed suicide. Seems like that would be a critical causal link. If a show about suicide is available on Netflix and no one is around to watch it...
- ggreer 7y ago
  They did not. The CDC data on suicides only contains statistical information, not identities. And even if the researchers did have the identities of everyone who committed suicide, I doubt Netflix would share account data with them. It would also be very noisy as many teens use their parents' Netflix accounts.
- boomboomsubban 7y ago
  No, all of the data is just taken from the standard CDC reports.

by ggreer 7y ago

sigstoat 7y ago

> I don't know how many ways they carved up the data before they found a segment of the population with increased suicide rate at the right time. and they still reported 9 different categorizations (overall/male/female * 3 age groups), and could only find one with a change. it isn't clear to me if they corrected for the multiple comparisons they did report.

ggreer 7y ago
Right, but that doesn't tell me how many other ways they carved the data up and didn't report it in the paper. Maybe they initially chose different age groups (10-22, 23-50, >50) and found no effect. Then they tweaked the categories and tada, controversial paper!
- sigstoat 7y ago
  > Right, but that doesn't tell me how many other ways they carved the data up and didn't report it in the paper. i understand that and i don't believe i contradicted you, rather i was emphasizing: they _admitted_ to 9 different carvings and only got 1 interesting one. what they did under the hood had to be even worse!

malandrew 7y ago

Maybe I missed it in the article, but did they check the Netflix viewing history of those that committed suicide. Seems like that would be a critical causal link. If a show about suicide is available on Netflix and no one is around to watch it...

ggreer 7y ago
They did not. The CDC data on suicides only contains statistical information, not identities. And even if the researchers did have the identities of everyone who committed suicide, I doubt Netflix would share account data with them. It would also be very noisy as many teens use their parents' Netflix accounts.
boomboomsubban 7y ago
No, all of the data is just taken from the standard CDC reports.