Analyst’s Crucible (Statistics)
Three Books That Will Challenge And Improve You
Analyst’s come from a variety of backgrounds, common undergraduate majors include computer science, engineering, mathematics, and business. Their disciplines provide solid exposure to things like statistics, logic, system, and the scientific method — although few cover everything and then only so deep. So analysts are forced to learn on the job or self-educate.
Self-Education
This article is about the latter. I am a big fan of self-education. I am a bigger fan of guided or enhanced self-education (more on that here). This short article is to provide you with a set of resources that can vastly improve your analytic prowess, your understanding of statistics, and your ability to make decisions. Best yet, it is as simple as reading a few books.
The Crucible
So why employ the Crucible reference? Any analyst reading these three books is going to confront a brutal reality. Most of our colleagues, and if we are truly honest — most of us, are doing it wrong! Or at a minimum — we are doing it lazy and that is resulting in misread outcomes and poor testing. Worse yet, especially for those handful of statistics majors who make their way to roles in analytics, executives across the corporate world are unknowingly (we hope) endorsing poor behavior.
This Has Been Going On From The Beginning
The first book in the trilogy that I will recommend was written in 1954, but Darrell Huff was simply following in the footsteps laid over the prior century. Notables from Mark Twain to Benjamin Disraeli took exception to the use (rather misuse) of statistics and the abuse of statistical ignorance. It is worth noting that neither Huff, Twain, nor Disraeli was a statistician. It is also interesting to add that, contrary to popular myth, none of them coined the phrase: “ There are three kinds of lies: lies, damned lies, and statistics.”
Huff’s book is a short read. It is 144 pages including illustrations (Irving Gels was quite renown in his day). It touches on sample sizes, visualization techniques, bias, and the evils of marketers (wink). It was an instant classic that did much to popularize statistics.
And It Is Not Getting Any Better
In recent decades, venerated institutions, journals, and even the peer-review process in general have come under grave scrutiny. Studies done by numerous third party institutions have conclude: most scientific studies submitted for publication are statistically invalid. In other words, it is not just business and marketing analysts — scientists, doctors, and researchers around the globe fail to utilize basic statistical standards. If brain surgeons and rocket scientists are taking short cuts, where does that leave the rest of us?
In 2015, Alex Reinhart published an excellent successor to Huff’s original. Reinhart needed an additional 20 pages to take on the scientific community. Arguably, it takes a little more to convince us that scientists can fail than it does to convince us that marketers can lie. Although if Reinhart’s book gets the recognition it deserves that may no longer be true 30–40 years from now…
If That Wasn’t Jarring Enough…
Statistical insight and understanding hasn’t been sitting still since Twain and Huff started poking fun at how misused it can be. Statistics, like all true science, continues to evolve and grow. Statistical processes in information age have only grown more complicated and confusing. This may be a good thing as ignorance in the use of t-tables and confidence intervals may be one of the most prevalent problems — talk about knowing enough to be dangerous.
Gary Smith required 304 pages to address the more modern issues with the application or misapplication of statistics. His take is still very approachable and will likely teach you a few things as well. As the subtitle indicates, Gary was very aware of the legacy of Darrell Huff making this third book the perfect closing piece in our set.
Are You Willing To Climb This Wall?
Three books. Just over 600 pages. And a challenge I have for all would-be analysts. Can you handle this analyst’s crucible? Can you handle the truth?
You could knock these out on a long international flight. But once you recognize the cracks in the system, things won’t be the same. People are not comfortable when you challenge their understanding. But the analytics discipline is definitely in need of A Few Good Men (and Women) to stand watch over our statistical methods and outcomes. Are you up to the challenge?
