“Most people use statistics like a drunk man uses a lamppost; more for support than illumination” – Andrew Lang
In 1969 I took my first college statistics course at the University of Nevada, Reno (UNR). The teacher was Gaming Hall of Famer Dr. William Eadington. It was, coincidentally, the first course Bill taught at UNR.
I stayed in touch with Bill for the rest of his life and even moved into his home for a bit when he became ill. One of the constants of Bill’s reality was that he was in absolute pain when reviewing my math and statistics work. While he was incredibly kind and generous, he did not have a lot of tolerance for less-than-perfect work, and I often specialized in less-than-perfect work.
I was once listening to an interview with the comedian Bob Newhart, who had been an accountant before changing his career trajectory to become an entertainer. He noted that he left accounting because his employers were troubled by his attitude of “that’s close enough.” That was often my mantra through an undergraduate, master’s, and five years of torture in a Ph.D. program—all in finance and economics. However, I did pick up a bit in all those years. One rather needs to.
I mention this because I just read through a research effort entitled “Diversity in the Gaming Workforce,” published by the American Gaming Association (AGA).
The AGA refers to this as “… an in-depth examination.” I do not see it that way, even by my “that’s close enough” standards.
Maybe I am wrong. The AGA does not allow one to see much under the hood, and maybe there are pages and pages of footnotes and annotations filling up a box somewhere that might make sense of their effort—but I can’t find it. I see a poor study design surrounded by an array of statements resting on a weak foundation.
As an aside, and by way of a nitpick, I was troubled by an early statement in the text: “Three in four Americans view our industry as having a diverse workforce…” I have no earthly idea as to the attribution for this statement. It just appears in a study that seems to be trying to measure diversity. But this is but a nitpick, and I have created greater sins in my past writings.
I was much more troubled by the methodology of the study.
At the end of 2022, the AGA offered its manufacturing and operating members the opportunity to turn over their EEO-01 submissions to a third-party accounting firm in an effort to generate “gaming industry” diversity metrics.
The EEO-01 submissions are required in accordance with Title VII of the Civil Rights Act of 1964 and the EEOC and track trends along a number of demographic variables. To repeat, each of the member firms of the AGA (which is not the “gaming industry”) in the manufacturing and operating segments was given the opportunity to turn these forms over to a third party doing the number crunching for the AGA.
One needs to understand that this is an example of convenience sampling, an invention having the advantage of being inexpensive. More specifically, it is known as voluntary response sampling, where each member of the population can opt to be in or out of the survey. The issue that is introduced by this type of sampling is called self-selection bias. The disadvantage of it is that it is also referred to as non-probability sampling, and as such, not much faith can be placed in it as a predictive tool. In other words, it is an inexpensive method and lacks predictive capabilities. That is the primary problem I am having.
There are others.
It appears that eight firms that the AGA classifies as manufacturers participated. When one looks at the AGA membership categories, no category is identified as “manufacturer.” I assume this is the group the AGA calls “diversified gaming suppliers.”
There are also 18 AGA member firms classified as operators that participated, broken up between commercial and tribal casino operators. One of only two footnotes in the study mentions that sovereign tribal nations are not required to make the EEO-01 submission. The footnote further suggests that “…several AGA tribal operator members voluntarily completed the forms for the purpose of this study.”
In case anyone else is curious, after some research, it appears that “several” can mean a number from 3 through 7. There are sampling issues here, especially if one believes that there are meaningful demographic disparities in the labor forces of tribal vs. commercial casinos (having worked in both the commercial and the tribal sectors during a long career, I think there is strong reason to believe there is). Moreover, in the text of the study, it is suggested that the tribal environments are different.
At this point, I reached my limit. I could only conclude that this was clearly less-than-perfect work.
As I have gotten older (and that is something that I have recently been doing quite well), I often ask a question when I am looking over a research project, and that question is: is this project helping me learn something? My answer here was no. It seemed instead that I was being sold something.
What is unfortunate about going to all of the effort to develop “…an in-depth examination” that does not appear to be an in-depth examination is that there are some much easier constructs one can use to provide insights on the diversity within the gaming industry. I have developed my “a picture speaks a thousand words” index. Here, one looks at a firm’s website and accesses the page that says “leadership” (or senior management, or executive team, or some such related thing), and if the pictures are predominantly men, the firm has a diversity issue. (Spoiler Alert: most all do).
A second indicator that I recently discovered is the “Bathroom Line Index.” At a recent sports betting conference on the East Coast, there was generally a long line of men waiting to use the Men’s Room, and there was no waiting for the Ladies’ Restroom. It was an amazing sight.
That pretty much sums up everything you need to know about the new gaming industry and diversity.