The Cult of Statistical Significance: Reader Recommended in Research Review’s Review

I have occasionally reviewed research here (shocking, I know, given this blog’s title), but seldom recommended anybody actually buy any. All that is about to change in this edge-of-your-seat suspenseful thriller story of love, lost, and learning (without the first 2). The work in question is Stephen T. Ziliak & Dierdre N. McCloskey’s The Cult of Statistical Significance: How the Standard Error Costs us Jobs, Justice, and Lives. Now, while I’m sure that most people are like me and will collect and analyze datasets for fun like philatelists do stamps or fantasy football league members do their teams, but for the small number who think that statistical/data analysis is not their favorite subject, and the tiny minority who might somehow find it as boring and impenetrable as the security barriers implemented to prevent our satellites, space shuttles, and so forth from hypothetical ways in which NASA might disrupt non-existent ecosystems in any and all galaxies (they seriously do this; we have a terrible track record when it comes to damaging Earth’s ecosystems and yet millions and millions go to protecting non-existent ecosystems using mechanisms we can’t know will work).

For the “minority” who don’t think statistics is just fun, there are precious few papers or books that exist which have even the slightest chance of being remotely interesting. I imagine Ziliak and McCloskey realized this, because even though their intended audience includes students looking to go into research, it is really more for the general public. In fact, it was often somewhat challenging for me to read because they so successfully stripped their book of jargon that there were places I was confused (“oomph” isn’t exactly a term one comes across in statistics literature).

The book addresses one of the most serious problems in scientific research: (null hypothesis) significance testing (NHST), which is both a set of statistical methods and a research design that is used in medicine, psychology, climate science, anthropology, linguistics, artificial intelligence, sociology, economics, law, etc. There are many hundred papers across fields and over decades that have attacked this ever-growing malignant tumor within the sciences, but it seems as if no matter how influential the academic or prestigious the journal, the only people who read it are the ones who already believe it (although there are signs this is changing somewhat).

There is one thing that can turn trends in the sciences around like nothing else, and it has nothing to do with scientific research (often for the best, such as how quickly public and privately funded eugenics programs as well as research were immediately abandoned after the holocaust, or how quickly homosexuality was dropped as a psychiatric diagnoses thanks to public outcry). That thing is public awareness, the kind that scales the ivory towers of academia and jerks academics away from the minutia we study to defend “big picture” issues with our practices.

Unfortunately, there are comparatively few fields that can readily arouse public interest, and statistical methods and their corresponding research designs don’t make the list- until now:

“No working scientist today knows much about Gosset, a brewer of the Guinness stout and the inventor of a good deal of modern statistics. The scruffy little Gosset, with his tall leather boots and a rucksack on his back, the heroic underdog in our story…For over two decades he quietly tried to eduate Fisher. But Fisher, our flawed villain…,” well, I don’t want to give too much of the plot away. The point is that instead of merely a technical argument, or even a simplified, informal version of a would-be technical argument, this book has incorporated history, fictional-like narrative in non-fiction, protagonists and antagonists, and real damages that has affected and indeed ENDED lives. Just go to Google Books or Amazon and look at the table of contents to see the breadth of the “stories” (true accounts that actually happened as a result of the problem addressed by the authors) the authors draw upon to make a subject that can make most people fall asleep and turn it into a drama without relying on sensationalist.

Finally, not only is it a sweeping narrative, a historical account, a glimpse behind the curtain of many fields that are immediately relevant to so many, but it manages somehow to communicate statistical methods and research methodology with so much accuracy given so little technical details or jargon I would kill to be able to simplify half as well.

Penrose’s The Road to Reality was a best-seller, despite the fact that after 7 or 8 out of the 30+ chapters it’s addressing mathematics that are taught at the graduate level and I’m pretty sure the only people who actually read it through were those that didn’t need to. Hoftstader’s Gödel, Escher, Bach: an Eternal Golden Braid was more successful in the sense that I think more of those who bought it actually read it, but few even realized (as Hofstader himself said) that the book’s topic was cognitive science. This books is far less technical, far more relevant, far more important in terms of relevancy to our society, and much more interesting.

Advertisements
This entry was posted in Mathematics, Statistics, The Language of Science: Why most scientists don't speak it, The Scientific Method and tagged , , , , , , , , , , , . Bookmark the permalink.

One Response to The Cult of Statistical Significance: Reader Recommended in Research Review’s Review

  1. Jeff says:

    Thanks for the review. Think I will get a copy and add it to my reading collection.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s