Words Matter In Poll Results

The two charts from a recent Gallup Poll show a real problem with some polls. If questions are not asked in a neutral way, without using words that have negative (or positive) connotations, the results of the poll can be seriously affected. And some political polls contain this error, as they attempt to put a candidate in a better light than is real (in the hope that a good poll will convince some voters to come on board).

Note that the two polls shown above are asking the same question. But while the first one couches the question in a more appealing way, the second uses the word "suicide" -- a word that has some negative connotations in this society (especially among many religious people). And the use of that word in the question lowers the percentage of people approving of a doctor ending a person's life by request by about 19 points. That's a huge difference. And that huge difference is not caused by different attitudes but by the different ways the question was asked.

The poll, done by the folks at the Gallup Poll, questioned 1,535 people between May 2nd and 7th. Those people were divided into two groups -- and one group was asked the question including the word "suicide", while the other group was asked the question avoiding use of that word. The margin of error was 3 points for the poll, and if the words used in each question didn't matter, the two groups should have numbers within that margin of error.

Before we get into another political season (with numerous polls being released), I just wanted to let you know that all polls are not the same. And you cannot know a poll's reliability until you know exactly what was asked and how it was asked. I know I'm probably preaching to the choir here, but I found this interesting and just thought I'd pass it along.