Friday, November 06, 2020
Stupid myth
This year as in 2016 there's some pointless "analysis" of why polls went wrong. The "analysis" won't change anything and isn't meant to change anything, just as "elections" aren't meant to change anything.
Expecting a poll to measure opinion is stupid. In general pollsters are paid to MAKE opinion, not to MEASURE opinion. One company or one party pays the pollster to create the impression that everyone shares one particular opinion. This shared opinion doesn't always agree with the company or party; sometimes the poll is designed to cause a rebellion against the "shared" view that will benefit the company or candidate. The other guy is winning! We need more funds so we can campaign harder!
There are some notable exceptions to this rule. Pew conducts several long-standing annual surveys of religion and opinion that ask the exact same questions every year. These surveys are more likely to measure opinion, though there's still room for altering the measurement by changes in sampling procedure.
Measurement ALWAYS requires a fixed standard that remains the same forever, and a fixed set of tools and methods that remain the same forever. When you redefine the standard, like the "social" "science" redefinitions of autism or the "virus" redefinitions of cases and immunity, you are NOT MEASURING ANYTHING AT ALL. You are making and implementing policy through the back door.
Labels: Metrology
¶ 6:06 AM
