Wednesday, June 24, 2015

Dishonest questionnaires and statistics

Assume that I had a political or ideological agenda that compelled me to prove that a good portion of the population thinks that murder is acceptable, and I want to prove this with a questionnaire. How would I do this?

If I were an honest person, I would pose the claim "murder is acceptable" and then provide two answers, "agree" and "disagree" for people to answer. Of course this method wouldn't give me the result I want, unless I'm so devious as to outright lie about the results. The vast majority (if not all) of the people will obviously answer "disagree". That won't do at all.

There are, however, other methods that can be used besides just outright lying about or distorting the results.

Suppose that instead the claim posed is "killing another person is acceptable" and now the range of possible answers is between 0 and 5, where 0 means "completely disagree" and 5 means "completely agree". What I won't tell the subjects is that the range is just an illusion, a diversion tactic, and that in reality I will be interpreting the results as there only being two answers: Zero and non-zero (with all non-zero answers meaning "I agree that murder is acceptable".)

This way I can trick people into giving me non-zero answers (which is what I want). When you give a range of possible answers between the two extremes, this introduces some nuance into the answer. After all, many people would agree that there may be some rare extreme circumstances where killing another person is acceptable, such as self-defense to save one's own life, or the police killing a criminal that's an imminent mortal danger to others, as the last resort. Such people would probably give a non-zero answer (most probably a 1, perhaps even 2).

But as said, this is all just a diversion tactic, a fraud. I will be interpreting all non-zero answers as "finds murder acceptable". This way I can make sensationalistic proclamations about the murderous nature of our society, based on actual questionnaires, with actual percentages. (Naturally I won't be proclaiming how I came up with those percentages.)

All this sounds really theoretical. The sad thing is that I pretty much repeated an actual recent "study" that employed this tactic. Not for acceptance of murder, but of acceptance of rape. This "study" found that something like 30% male college students thought that forcing a woman to have sex is acceptable.

Egregiously, if you actually dug the details of how they made the "study", it was exactly as I described above: A claim more ambiguous than simply "rape is acceptable", and a range of answers from 0 to 5, and the "study" interpreting all non-zero answers as "finds it acceptable".

I will perhaps never understand why social justice warriors are so eager to inflate numbers and distort statistics. It's dishonest to the core. And what's the purpose of it?

No comments:

Post a Comment