Here is link to the COMPARE project, an incredible, though depressing, project designed to check if clinical trials are following their preregistered analysis plans when published. Pre-registration is a great technique to help draw a bright line between planned and…
Introduction to the New Statistics
The online community and blog for ITNS
Month: August 2016
Draw your own data tool
Sometimes it is nice to be able to make up a set of data to explore. Here is a cool tool that makes it easy to craft your own data set: just draw the data and you will instantly see…
Criminal justice and the perils of regression analysis
Regression analysis is incredible–it can literally help us predict the future based on patterns observed in the past. There are many pitfalls, however, to using regression. First, the correlations observed in the past may not apply to new cases or…
Tagged with: Regression
https://science.sciencemag.org/content/367/6475/240?utm_campaign=toc_sci-mag_2020-01-16&et_rid=17172836&et_cid=3167079
“In an era when every restaurant is obliged to publish on its front door the hygiene rating in their kitchen, we're seriously not saying whether a trial that costs millions of dollars has broken the law [...] by failing to report its results?”@bengoldacre
More evidence that superstition does not strongly influence performance.
https://econtent.hogrefe.com/doi/abs/10.1024/1010-0652/a000263
Still, the idea won't die. The original study was cited 29 times last year (2019) vs. 5 cites for the replications I conducted with @Fernie103 showing negligible effect.
Sensible reporting guidelines for exp bio: show the data, make confirm vs. exploratory clear, emphasize effect size and uncertainty. Good resource.
Two colleagues and I just published an editorial/review for how to present data and stats in experimental biology journals. It explains new guidelines adopted by three American pharmacology journals.
http://molpharm.aspetjournals.org/content/molpharm/97/1/49.full.pdf
Polisci types: Beyond voting, what are the most effective things I can do towards the 2020 elections? ($ to a pac? A candidate? Volunteer to canvas? Phone bank?)
What is actually helpful?
Never cared so much before. Not asking for a friend.
(@BrendanNyhan?)
Doing more with confidence intervals. https://replicationnetwork.com/2020/01/10/doing-more-with-confidence-intervals/
New Stats blog: Bushfires and Open Science: A View From Australia https://thenewstatistics.com/itns/2020/01/09/bushfires-and-open-science-a-view-from-australia/
Cheers to the guy who made psychological science—the journal and the field—better. https://twitter.com/dstephenlindsay/status/1212063000951021573
My last day as Editor of Psych Science. Whew! Best wishes for new Editor Patricia Bauer!
New publication from our research group in @NatureHumBehav showing that meta-analyses substantially overestimate effect sizes and that commonly used methods for bias adjustment do not substantively improve results: https://www.nature.com/articles/s41562-019-0787-z
Top
Recent Comments