Numbers mislead
Survey data can be skewed
Seventy percent of Americans think government is doing a bad job. Fifty percent of voters say that political parties are doing what they should. Sixty percent of consumers believe that corporations simply want to increase profits.
Surveys are fascinating things, misleading and full of skewed information. After all, 100 percent of the people surveyed for this article believe that surveys are always erroneous. Yes, 100 percent.
I love seeing commercials and ads where percentages are touted for whatever is being advertised. All that is necessary is the bewitching “four out of five dentists believe …” and we are hooked. Irrelevant where those dentists are. Irrelevant if they are all in the same office and supplied massive amounts of free samples of the product they so gushingly promote. Irrelevant if they are even real dentists. Don’t we see the fine print: “compensated spokesperson” or “actor portrayal”? Not even real people, let alone real dentists.
Demographics and makeup of a survey are usually left out of the results. Watching CNN or any news show and it is so annoyingly repeated, “Eighty percent of Americans have lost faith in all except their local football team.” As if all 300 million-plus of us were surveyed. Extending the results of perhaps 3,000 people being asked misleading or leading questions to represent any number of people is wrong, plain and simple.
A class I took way back in my undergraduate days, the Stone Age of the early ’80s, drilled into me an appreciation for test norms, reliability and validity data. This class studied academic tests, intelligence tests, math and reading tests used in diagnosing learning disabilities and other handicapping conditions. The first thing I always looked for was the sample population—where it was selected
from, ages, other demographic aspects, and populations on which the test were used. The same with surveys.
I realized then that data could be skewed and manipulated to portray a variety of outcomes. A survey based on the questioning of maybe 2,000 people is, to me, in no representative of the nation as a whole. Asking “potential voters” who they prefer in an election does no good at all. Remember, these are “potential voters,” not even people who have decided if they are going to vote. And please don’t get me started on how the choices of 50 percent of those who bother to vote, which is roughly 50 percent of the electorate as a whole, somehow signifies a momentous landslide or national mandate.
Surveys, I suppose, some find useful. One hundred percent of those surveyed believe that some surveys do serve some purpose. Of course, the sample size for that survey was one. Irrelevant, as the main attraction isn’t actual numbers but that all-important percentage.
I shudder each time I see or read that “80 percent of Americans believe …” when maybe only a couple thousand were questioned. Wouldn’t it be nice if each time a survey was quoted that it was truly quoted? “Eighty percent of the incredibly small sample population which in no way reflects the actual population but we have to sound important so we quote it …”
Oh well. A person can dream. And in conclusion I will simply state, “One hundred percent of those surveyed believe that, “As always, slow down and grill.”