4 ms·
I can't believe the paper has no scatterplot. Maybe the Internet can help. Anyone want to make one?
by pg 14y ago
I can't believe the paper has no scatterplot. Maybe the Internet can help. Anyone want to make one?
- zt 14y agoI would love to make a scatterplot and also run a few other statistical models on the data. Unfortunately, hitlab.org is not up and, more generally, medical studies have a much lower data replication standard than some of the social science. Neither the corresponding author's email--author@hitlab.org--nor his public academic email address seem to work...
- nswanberg 14y agoGoogle Docs will OCR images, so here is a start on the raw data: https://docs.google.com/document/d/1moZ9bh9d2HamF-fEYUkqAROBt_tzUB6DAEDn8MHQoJI/edit?usp=sharing https://docs.google.com/document/d/1moZ9bh9d2HamF-fEYUkqAROB...
- nswanberg 14y agoAnd here's a scatterplot alongside the data. Feel free to edit or copy as desired: https://docs.google.com/spreadsheet/ccc?key=0AiBnrQZn4821dHV3WGFMRWJTbTRvY0Z3QmdmdGVSY2c&usp=sharing https://docs.google.com/spreadsheet/ccc?key=0AiBnrQZn4821dHV...
- graphene 14y agohttp://imgur.com/33mYSic http://imgur.com/33mYSic This is from the paper's table 1, not having done any of the statistical analysis to get rid of the confounding factors (age of facebook page, size of hospital, etc).
- pg 14y agoThanks! The trend is not exactly conspicuous.
- graphene 14y agoto say the least.. When plotted on a linear "likes" scale, the point with the most likes stands out more, and would probably cause the linear fit to have a negative trend, because its mortality rate is well below average. However, I don't really think it's justified to draw conclusions from this either way without taking into account even hospital size and fb page age.