Sometimes you read a book or paper that’s so good you just have to share it.
One paper I recommend to all my trainees is this David Freedman piece. The central theme is that all the sophisticated analysis in the world cannot replace thinking long and hard about the problem at hand. One quote I like in particular, referring to Snow’s work on cholera:
“At every turn, he anchored his argument in stubborn fact. And he exposed his theory to harsh tests in a variety of settings.” p 304.
Easily the best free resource I’ve ever used: https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/ Read it and save yourself a few thousand dollars on an MPH course at Harvard!
Probably known by some of you already but I really have to share McElreath Statistical Rethinking as the best Bayesian Introduction for applied researchers and probably the one from where I learnt the most.
This book by Royston and Lambert is a fantastic introduction/guide to survival modelling beyond traditional Kaplan-Meier and Cox proportional hazards models:
- The now rather famous paper by John Ioannidis “Why most published research findings are false” is still worth a read for all working in science.
- A recent detailled but very readable explanation about the flaws of NHST is given by Szucs et al. (2017).
- I found the two books by Jos Twisk: Applied Multilevel Analysis and Applied Longitudinal Data Analysis for Epidemiology very readable, especially for beginners.
- Another great book on longitudinal modelling is the book by Judith Singer: Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence
if you read Ioannidis you must also read the rebuttals; the responses are mostly lost in the 1000s citations from popular press etc [edit: example: response to Ioannidis]
This is a great article. It brings to light some of the ethical issues in drug development. Even more relevant now, given the ‘right to try’ legislation. This type of info is important to be embraced by the media and general public.
I am interested in how you all would respond to this interview with John Ioannidis.
Hopefully the link posts correctly.