The dangers of using partial statistics

Updated: Oct 5, 2020

The debate between Jordan Peterson and Cathy Newman on the issue of the gender pay gap was surely one of the most interesting events of the past three years. Numerous and intense exchanges between the two would become the subject matter of countless internet memes, and Jordan Peterson would arise to a level of great notoriety. However, there was one exchange that I find was of particular importance in the whole interview, more than others. Allow me to draw some quotes and add some personal commentary in brackets.

That portion, around the 5-minute mark, saw Cathy Newman push the point:

"What I want to put to you is that here in the U.K., for example, the gender pay gap stands at just under 9%. You've got women at the BBC recently saying that the broadcaster is illegally paying them less than men to do the same job. You've got only 7 women running the top Fortune 100 companies. So it seems to a lot of women that they are still being dominated and excluded."

To this, Jordan Peterson replied,

"It does seem that way, but multi-varied analysis of the pay gap indicates that it doesn't exist. (In the sense that was implied, as being a result of gender only. ) There are multiple reasons for that. One of them is gender, but it is not the only reason. (For a pay gap.) Like, if you're a social scientist worth your salt, you never do a uni-varied analysis. Like you say, women, in aggregate, are paid less than men. Ok, well, then we break it down by age, we break it down by occupation, we break it down by interest, we break it down by personality. "

"I'm saying that the claim that the wage gap is only due to sex is wrong, and it is wrong. There's no doubt about that: the multi-varied analysis have been done. So I could give you an example. Ok, there's a personality trait known as agreeableness. Agreeable people are compassionate and polite, and agreeable people get paid less than disagreeable people for the same job. (Because agreeable people, those afraid of controversy and risk of a loss of favor tend to refrain from asking for a raise.) Women are (statistically) more agreeable than men. "

"That's one component (trait agreeableness) of a multi-varied equation that predicts salary. It accounts for maybe 5% of the variance; you need about another 18 factors. One of which is gender. And there is prejudice, there's no doubt about that, but it accounts for a much smaller proportion of the variants in the pay gap than the radical feminists claim. "

Cathy asked Jordan Peterson elsewhere in the video:

"What do you mean by that, that equality of outcome is undesirable?"

He explained:

"Well, men and women won't sort themselves into the same categories if you leave them alone to do it of their own accord. We've already seen that in Scandinavia. It's twenty to one, female nurses to male nurses, something like that, it might not be quite that extreme. And approximately the same male engineers to female engineers. And that's a consequence of the free choice of men and women in the societies that have gone farther than any other societies to make gender equality the purpose of the law. Those are ineradicable differences. You can eradicate them (only unnaturally) with tremendous social pressure and tyranny. (I.e., in accordance with chapter 4 of the Communist Manifesto, "forcible overthrow of all existing social conditions.) But if you leave men and women to make their own choices, you will not get equal outcome." (And of course, some of these jobs that males tend to prefer more than females are higher-paying not because of the fact that it is a male doing it, but because of the fact of the value of the position and product produced. Machine repairmen servicing equipment in the manufacturing industries would be a great example. Those are high-paying jobs, but they are quite rugged, dirty, and physically demanding, and women tend to avoid those in favor of a desk job, which is seen to be cleaner and more delicate. Not many people are holding women back from becoming machine repairpersons, they simply tend to not pursue those careers by their own preference.)

For me, that debate was a wake-up call. Without considering a multitude of factors, anyone could be siphoned into an agenda or political movement which may not be operating in accordance with a complete picture drawn from a comprehensive body of data. In the same way that a 9% pay gap doesn't tell much of its own, but can give a very strong appearance when accompanying the right narrative, we surely have other issues like this.

For example, I recall one interesting event that occurred when I worked with the Hillman Group, ltd, in the warehouse that they formerly had at 55 Milne Avenue. One fateful day, the supervisor Robert, asked me into the office. Standing at one end was one of the managers, and also sitting there on a desk with a laptop open was Caroline, one of the higher-ranked office workers who would follow-up on customer order progress and also the accountability section of the WM-Oracle program. I had just spent 1 day training four-five different individuals who each arrived at different times (some after I already commenced the training) for different reasons (not necessarily negligence or intentional lateness), with different learning abilities and different needs for learning. For instance, one man named Ali was capable but was somewhat of a slower learner for the RF scanner basics. He could learn them, but he needed more one on one training and time. He had other strengths, though; he was very observant of physical discrepancies and would catch things others might miss. One of the others, Devante was a quick learner for programs like WM and could master the usage of the RF scanner and its software swiftly and thus operate independently sooner. This made training them all very complicated! We also had a shortage of RF scanners available at the time, and it took a long time to get individual scanners into the hands of each person; for about 2-3 hours, they each spent time sharing one scanner and one account, practicing picking stock, and by the time they each had a scanner of their own another 2 hours. Suddenly, Robert, the Supervisor, asked me, "Erik, I want to know who you think is worthy of keeping in the company."

This was a very awkward situation for me. To be honest, I had worked in the warehouse for about two and a half years, and I knew there was absolutely no way to gauge someone's potential growth and usefulness in one day. This environment was not original to the warehouse; when I started working there, the RF scanners were not yet implemented, and orders were done using pen and paper, and at that time, no one thought to analyze workers this way and give them a single day to prove themselves. After about a year, the scanner system and WM launched, and now office workers like Caroline had the ability to remotely spy-out various numerical indicators pertaining to employee performance, which they never had access to before. Items picked, Items picked per minute, the time spent between picking one item, and the next would be automatically logged on the system and reported to the appropriate office workers.

Caroline argued for at least two of these men to be let go. She pointed to the monitor with total confidence at just two pieces of data: "Lines picked" and "Time between picks." Those with over 100 lines, for instance, and less time between picks were more favored than those who had 35-40 and more time between picks. Thankfully, I was somewhat knowledgeable with administrator functions of WM, and I knew some more of the metrics which were being ignored, and also had access to them since I had a SuperUser account. I argued, "Yes, Ondoua (one of the men I was training), for instance, has this many lines, and it does not look good, but you cannot tell the full story with only that little data alone. Take a look at units picked and the picture changes. The ones picking 35-40 lines in 2-3 hours are carrying more units, and it means the total weight of the product they were transferring was greater, but it is allocated as one line in this case though there are more products and heavier lifting involved, which slows down the process. In that case, the one who gets 100 lines in 2 hours is like somebody who wins a lottery because they happened to get a pick-ticket with a bunch of small and light items each counted as a line; and if Ondoua had a pick-ticket of this nature, he too might well of broken 100 lines for all we know." (At this time, the system was not well-refined at the time, and we had instances where 1 line could literally be over ten cases, each weighing 40-50 pounds, which would fill a good portion of a skid; each would have to be picked from the Active Shelves and moved on a cart to the appropriate location. Sometimes the carts couldn't handle the weight, and multiple trips were needed. In other cases, 1 line could be one pack of 100 screws that weighed less than 2 lbs., and the picker could clear their entire pick-ticket before needing to return to the designated area in a very short time. Caroline was treating these as if they were exactly the same.) So, Caroline and I got into a bit of a heated debate, but the manager agreed with me on a lot of points, and all the employees were kept, and all of them would prove themselves valuable overtime.

Caroline, at this moment, had reminded me of Cathy Newman. She was just looking at just one portion of the data and formulating a narrative, but not at the full story. And there were other factors from what I just mentioned. The point I argued was that it would take over a week to really get a good idea of what the employee's potential was, using more complete data. In this case, had I not stepped in, some of these men would have possibly lost their job.

I wonder how many instances in life do we have where we are dealing with people pushing agendas at the expense and livelihood of other people who are relying on one small piece of data that they overemphasize? It has been said, "the numbers do not lie." Unfortunately, people do lie, and people often present numbers out of context to move people according to their own desires and say things without thinking.

17 views0 comments