The Thing About Statistics

Helping you see what you wanna see.

To make surveys and statistical research instruments more accurate we are taught to insure that nothing about the conditions of the test encourages one group to participate or not participate more than others. I took two graduate level statistics courses to learn that. Of course, the other guys who took those graduate statistics courses know this and, as I did, learned all the sneaky ways there are to make a study or survey's data say what the researcher wanted it to. 

 

Knowing how to game the system is a very valuable skill in the political marketplace. So long as the researcher is careful to make these sorts of tricks very hard to find for the normal person untutored in the ways of statistical analysis, statisticians can make very good money designing studies that skew the results in exactly the way the client wants it skewed in order to make the study say what the client wanted it to say. 

Why do you think the polls had Hillary winning by a landslide in 2016? First it would convince the voters that her victory would be inevitable and people would either vote with the perceived herd or not bother to vote at all since there was no chance Trump would win.

The other reason the polls favored Hillary was that by selecting a data sample from large urban areas, there would naturally be more left-leaning respondents in the sample. It's a little trick called selective sampling. It was profitable to do so, because the people paying for those polls (mostly the liberal media) don't like to buy studies that don't show progressives winning. Conservative media is a much smaller niche market.

That's why so few polls tend to show conservatives ahead, especially in the early stages of an election and right at the end. Often they feel some guilt in the middle and publish a more accurate data set, but go back to the more profitable selective sampling and rigged questions that skew the results their way.

The US census is a golden opportunity every ten years to bend statistics to the service of the deep state. The federal and state bureaucracy has a natural inclination to prefer left-leaning big government politicians and therefore are highly motivated to tell them what they want to here and by extension, tell the public at large what big government liberals want the public to hear. Government employees keep their jobs where they sit at desks all week accomplishing little and getting paid on Friday by telling the people who vote to enlarge government bureaucracy what they want to hear.

Just giving out my little warning here. Don't put too much stake in polls in a race that's close to the wire. Herd instinct will pull many voters toward the side that seems to be winning. One of the reason Trump likes big rallies and parades is that the visuals there help overcome the negative polls and tend to neutralize the media bias with voters as they stand in the booth punching out those chads!

© 2020 by Tom King