I got 100 bucks for anyone from @jhubiostat that can explain what was actually done in this paper!
#statstwitterhttps://twitter.com/JohnsHopkinsSPH/status/1220003201165996033 …
-
Show this thread
-
Replying to @statsepi @jhubiostat
David Manheim Retweeted Johns Hopkins Bloomberg School of Public Health
Oh! I know this one - it's HARKing. And it's why we need to preregister studies in epidemiology the same way we already require in medicine. https://journals.sagepub.com/doi/abs/10.1207/s15327957pspr0203_4 …https://twitter.com/JohnsHopkinsSPH/status/1220003201165996033 …
David Manheim added,
Johns Hopkins Bloomberg School of Public HealthVerified account @JohnsHopkinsSPH"The largest apparent protective effect was found for children who had a household pet dog at birth or were first exposed after birth but before age 3" https://hub.jhu.edu/2020/01/06/dog-schizophrenia-risk-2499-em1-art1-rea-health/ …1 reply 0 retweets 1 like -
It's even worse than that. It's not just HARKing. The authors really seem to struggle with the connection between what they did and what they think they did.
1 reply 0 retweets 6 likes -
Replying to @ADAlthousePhD @davidmanheim and
For example, consider this passage in the Discussion: "Cox proportional hazard analysis indicated that exposure to a pet dog during the first 12 years of life was associated with an approximately 25% decreased hazard of having a subsequent schizophrenia diagnosis (Fig 1)."
2 replies 0 retweets 2 likes -
Replying to @ADAlthousePhD @davidmanheim and
Now go look at Figure 1. It's a KM curve of "time to getting a dog" with the groups being schizo, bipolar, and control. So a Cox model using this data would (presumably) not be "the hazard of having a schizo diagnosis for people with vs without dog" but rather...
2 replies 0 retweets 3 likes -
Replying to @ADAlthousePhD @davidmanheim and
...it would be "the hazard of getting a dog for people who are eventually diagnosed with schizo versus not eventually diagnosed with schizo"
1 reply 0 retweets 4 likes -
Replying to @ADAlthousePhD @davidmanheim and
The charitable version of events, assuming the analysis was even done correctly, still is that the authors really struggle to understand the connection between the model they used and the inference that is appropriate to draw from it.
3 replies 0 retweets 8 likes -
Extremely charitable.
1 reply 0 retweets 1 like -
Replying to @statsepi @ADAlthousePhD and
Looking at this in more depth, I'm even less enthused. The control group was a bunch of people picked from the same geographical area, and they report the results of a multivariate logistic regression model as relative risks
1 reply 0 retweets 4 likes
Oh shit, just realized - they analyzed cross sectional data longitudinally. This was based on the reported age they were when their family got a pet, not a yearly data collection
-
-
David Manheim Retweeted Johns Hopkins Bloomberg School of Public Health
New study, rephrased: Less depressed people more likely to remember having a pet as a young child.https://twitter.com/JohnsHopkinsSPH/status/1220003201165996033 …
David Manheim added,
Johns Hopkins Bloomberg School of Public HealthVerified account @JohnsHopkinsSPH"The largest apparent protective effect was found for children who had a household pet dog at birth or were first exposed after birth but before age 3" https://hub.jhu.edu/2020/01/06/dog-schizophrenia-risk-2499-em1-art1-rea-health/ …0 replies 0 retweets 4 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
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.