The Cult of Statistical Significance

User avatar
quantropy
Posts: 131
Joined: Sat Jan 28, 2017 10:38 am

The Cult of Statistical Significance

Postby quantropy » Fri Aug 17, 2018 12:42 pm

Why I looked at this book
In the quest to make sure that results are credible, statistics is becoming more and more important in science, and in particular in the medical sciences. But somehow it doesn't seem to work. Clinical trials should provide reasonable certitude as to the effectiveness and safety of a treatment, but it seems to often happen that a few years after the trial a fatal flaw is found. And trying to tighten up the testing looks likely to exclude life-saving treatments. This book claims that the quest for statistical significance has been a great mistake. I'm hoping that this will help to explain current flaws in medical and simliar trials, and tell how they might be done better in the future.

Tags:

User avatar
quantropy
Posts: 131
Joined: Sat Jan 28, 2017 10:38 am

Re: The Cult of Statistical Significance

Postby quantropy » Wed Aug 22, 2018 11:26 am

First Impressions
The introduction puts forward a good case for the claim that statistics has lost it's way, and that the sole focus on signficance testing is a mistake. The authors argue that somehow precision has become more important than oomph. We are introduced to William Gosset (the Student of Student's t-test) who worked out how to do testing properly, and then to the much better known Karl Pearson and Ronald Fisher, who let us astray. After a while though I felt that the material in the introduction began to get repetitive and I hope that when I get to the main part of the book it is more engaging.


Return to “Economics”

Who is online

Users browsing this forum: Antoniodmy and 1 guest