Since the days of the EPA secondhand smoke risk assessment, JunkScience.com has had contempt for the bogus statistical technique of meta-analysis. We have also taken the lead in exposing as fraud claims that ambient air quality kills people. JunkScience.com friend Stan Young has just published a new paper dismantling a major 2012 air quality study (headlined below) as the product of publication bias and p-hacking, thus exposing its meta analysis methodology as just a bunch of junk.
The Department of Interior has just issued a science transparency directive to its staff (see below). DoI’s intention is to enshrine science transparency via rulemaking. Another YUGE win for us.
I’ve always suspected that Stanford University professor John Ioannidis was only posing as a science reformer. His commentary in PLoS against the EPA science transparency rulemaking validates that.
Now finally available in PDF format, here is my 1994 report on science policy that the Clinton Administration tried to suppress. You can use it to comment on the just proposed EPA science transparency rule.
Get your comments to EPA before May 30, 2018.
P-hacking is where researchers collect or select data and statistical analysis until nonsignifiant results become significant.
‘Science’ is facing a replication crisis.
Here is a 1987 article from the great Ernst Wynder (1922-1999) author of the first large-scale study to link smoking with lung cancer. In the 1987 article, Wynder discusses the problems of drawing causal connections based on weak association epidemiology. Readers of this page know that weak association epidemiology has long been abused by government regulators, especially the EPA (Read “Scare Pollution” for a thorough treatment of this point). That’s why JunkScience.com has petitioned President Trump to issue standards for the use of epidemiology by regulatory agencies. This article is the introductory one to a 1985 workshop on weak associations published in Preventative Medicine in 1987.
As long as I’ve been involved in federal regulatory issues (since 1990), regulators have used junk epidemiology to justify overregulation. For the first time ever, we have an administration that is committed to stopping overregulation. So JunkScience.com petitioned the Trump administration today to stop the misuse and abuse of epidemiology by issuing epidemiologic standards for federal agencies.
So like, JunkScience has been been saying this for 21+ years. But whatever…
A great example of differential misclassification bias that was a criticism of the secondhand smoke epidemiology — i.e., smokers pretending they were nonsmokers. Here’s a quantification of the effect.