During my 20-year Information Technology career, statistics were an invaluable tool for solving user issues. In one example, I ran an SQL analysis on a program that revealed one of four database files was accessed the most often, yet it was the third check in a conditional decision chain. By changing the order of the decision chain to make it the first check on the list, the program ran faster. In another example as a systems administrator, I discovered that overall on-line response time was lengthened in inverse proportion to the amount of background jobs executed at a certain time in the business day. By moving some background jobs to execute during at a different time, the response time improved noticably.
Statistics does not necessarily mean truth, however. My approach in solving the problems in the above examples were correct only because the statistical data provided clearly showed the correlation between cause and effect. There are cases where the percentage of probability, though high, can still be wrong based on the sample data provided.
Let’s pretend that for one hour I did a count of all the people entering the Tim Horton’s where I’m writing this post right now, and found that 77% were of Asian ancestry. That does not mean each day and every time, in this Tim Horton’s, slightly more than three of every four customers are of Asian ancestry. Perhaps a few buses full of Asian tourists suddenly stopped in front of this Tim Horton’s to get a doughnut or a coffee, meaning random chance just took a steaming dump on my analysis. In fact, there’s nothing on the display case at this store that would attract Asian patrons over other ethnicities. In this case, my sample data did not accurately measure the correlation between cause and effect, even though the statistical probability was high enough to lead to the incorrect assumption
It is for this reason I cringe when I read career articles like this one that state a certain type of worker is going to behave in certain way based on age, colour, or whatever. In the case of the employment study I’ve linked, it’s assumed by the statistical data in the report data that Millennials (those under the age of 30) are untrustworthy.
Studies like this reminds me of the racial profiling done by certain levels of law enforcement in order to determine where types of crime were coming from. The problem with that type of profiling, be it for employment measurement or for catching bad guys (and gals), is there might be other reasons why things are happening as they are, yet are not being considered as a possibility.
The report’s conclusion is laughable. Does this mean the moment someone turns 31, they are now automatically trustworthy, as if a switch is flipped? It’s as insane as assuming people commit crimes because of skin colour or the faith they follow. Yet here we have a study that some employers no doubt will read and as a result practice a form of discrimination that is not only unfair, it is something a job applicant who falls into a particular statistical percentage will have a hard time overcoming.
I believe each of us has a personal ledger we are individually responsible for, for all the past actions and decisions made. We can’t blame others if our ledger is running in the red (morally speaking), but at the same time each of us shouldn’t be blamed for the actions of others. When it comes to statistics used in judging someone, the number 1 — that one person alone — is all that should be considered.
Thanks for reading!