While looking for some important stuff on the Harris Poll website (okay, incidence of sexual encounters with garden furniture), I couldn’t help but notice this absolute gem of a poll results title:
It’s just like the old joke of the statistician asking the client “What would you *like * the survey results to be?” and having a means to mash the data around to fit.
For several years I was an aptitude testing specialist with a government agency. When we issued test scores, the scores were reported twice: the achieved score and the score plus the standard error of measurement. More than once, people called me to ask whether these acknowledged “errors” invalidated the test scores.
