Police Psychology | Lies, Damned Lies and Statistics
by Gary S. Aumiller, Ph.D. ABPP
I can’t wait to see the polls this week. Two weeks ago Trump was behind by 10 points, earlier last week by 5 points, the end of last week by 2 points and yesterday he was ahead by 3 points in the states that mattered. Yeah right! I can’t wait to see what kind of fantasy the news wants us to buy this week. Not that these really aren’t the polls, but the methods and presentation do not seem very accurate anymore. In fact, statistics do not seem accurate anymore in general. You can’t trust them. It brings back the old quote from Mark Twain “There are three types of lies: lies, damned lies and statistics.”
In 2001, a student and professors dealt with rumors that Greek Hospitals were doing large number of appendix operations on Albanian citizens. It was not reported in the statistics. So they studied the rumors in six hospitals and in fact an Albanian was 3 times more likely to have a healthy appendix taken out of them than a Greek citizen. Three times more likely to undergo an operation that wasn’t needed! It was reported in an article called “Lies, Damned Lies and Medical Science.” What would a “lies, damned lies and policing” look like?
It has been said that 92% of statistics are made up on the spot. Sound a little high, maybe it is more like 76% of statistic are made up on the spot. Actually, I have seen from 26% to 92% when talking about what is made up on the spot – sort of evidence that the premise is true whatever the number is. So let me just point to four ways statistics can lie to you by looking at some of the myths of policing.
The Wrong Presentation
“A police officer is ten times more likely to kill himself than be shot and killed in the line of duty.” Therefore, with a little logic, the suicide rate is very high with police officers and they are prone to depression and can become mentally unstable. WRONG, mostly at least. The initial statement is probably right for most professions, except maybe bridge workers or skyscraper welders. Very few people get shot and killed, or even die in the act of doing their work, than kill themselves. It is a meaningless statistic. When officers were held consistent for sex and age their rate is a little lower than the general population. Mike Aamodt showed that in a study for an FBI conference bringing together the top people in the field of suicide in law enforcement. I showed the same type of data, but not as conclusive as Dr. Aamodt. Now one suicide is too many as far as I am concerned, but generally cops are pretty stable individuals. We should continue to work with them on lowering the rate, but realize we are working with healthy people.
But, the statistic was presented in a fashion that make it totally worthless. It is the presentation that makes the statistic worthless, in fact it is not even a statistic at all. For years it was backed up by showing the general population rate of all people with a lower rate, but very few five year olds kill themselves. So they destroy the value of the statistic by comparing two groups that are meaningless. It’s only when you hold the statistic for the age of officers does it pop out that the police rate is not much different. Dr. Aamodt is better than 99.9% of researchers in understanding statistics, I know, he told me so. (And it probably is true!)
The Wrong Statistic
If you look at the average income in the area I live it will say that $72,000 is the average income for the area. Look closer and about 67% are below average. There is no limit on the high end of income, but there is definitely a bottom limit, therefore the whole average statistic is skewed, meaning it’s worthless. Many people feel below average but are really above the median income for the area they are in, which for my area is around $50,000. If you look in policing, the average number of citizen complaints is often used to show that a person does their job without causing problems to the citizenry. Again the wrong statistic as citizen complaints are highly correlated to area worked, number of arrests, and the type of job a person is doing. Telling when a person falls in the average range is a meaningless statistic because it is the wrong statistic.
Whether it is an average, or a multiple regression, people tend to know when they are using the wrong statistic. This is a way to purposely present the data in a misleading fashion to make you think a certain way.
Here is one statistic I use to make people less judgmental that you can’t argue – “Half the world is below average in intelligence so don’t be so judgmental.” Stop and think about it.
The problem is that when many people see great big correlations and they start to believe in causation. Fisher in the 1950’s came up with a phenomena called the “Twisted Pear” that has pretty much been ignored. He said high correlations are often a result of part of a curve when really after that there is little correlation at all. Yet because part of the curve has such high correlation, the overall correlations look pretty good, when in fact it is not. On a graph it looks like a “twisted pear,” thus the name. So, whereas a test might predict whether a person can do a job or not at the lowest levels, it does not predict gradients of performance of the job. “Success predicts success” is usually a better indicator than a test score. When correlation gets confused with causation all kind of problems occur, and that is a general problem around the world.
Validity of the Face
Finally, we were taught as Ph.D.’s that there is many different types of validity – internal, external, predictive, test, content, construct, criterion, convergent, discriminate and face validity. I don’t really want to go into all of them and you most likely don’t want me to. But face validity is the quality of the validity that looks good and makes sense, and also has the least meaning. There’s not statistics for it, just whether it look like it will work. These days that is all that seems to matter anymore is the face validity. Does it look good, make me look good and does it shut people up. Science goes to the wayside. I think the Black Life Matters Movement falls in this area, as the honest statistics and the situations don’t bear out the premise that cops shouldn’t be more nervous around persons appear out of the mainstream and carrying guns.
As the quote “There are three kinds of lies: lies, damned lies and statistics” is attributed to Twain, but he attributed it to a British Prime Minister Benjamin Disraeli. The problem is no one ever heard Disraeli say it and he didn’t write it, in fact the phrase never appeared anywhere until well after his death. At least a dozen others are attributed this quote. I only knew of Mark Twain, so I attribute it to him. If you don’t like it – make up your own person to attribute it to, or say it was you. That seems to be what is being done in the polls these days.
Site Editor: Gary S. Aumiller, Ph.D. ABPP
Please go to https://www.kickstarter.com/projects/thinbluemind/the-thin-blue-mind-video-courses-and-inside-police and read the campaign
Please share this article from down below.
Please join the email list on the top of the sidebar and you can get these sent to your email. Also follow me on Twitter for other articles and ideas,
Come back regularly for more updated articles on police psychology