The Washington Post released a poll today that dealt with health care along with several other issues. Unfortunately, Ceci Connolly and John Cohen, who wrote the article about the poll, failed to properly read it.
The Washington Post story stated, "In the new Post-ABC poll, 62 percent support the general concept [of a public option], but when respondents were told that meant some insurers would go out of business, support dropped sharply, to 37 percent."
That would be an interesting statistic. The problem is that it is incorrect and does not exist. What the poll actually asked respondents was: “21a. (IF SUPPORT) What if having the government create a new health insurance plan made many private health insurers go out of business because they could not compete? In that case would you support or oppose creating a government-run health insurance plan?”
The use of word many instead of some is very important. If you think I'm being a semantic stickler, let me ask you a question: Would you be less inclined to go to Baskin Robbins if I told you they were planning to eliminate some of their 31 flavors, or if I told you they were planning to eliminate many of their 31 flavors?
“Some” would make me think they are dropping the 3-4 least popular flavors. If I heard “many,” I would think they are eliminating around 14-19 flavors.
Polling is a very delicate business. Small changes in phrasing will have a dramatic impact on polling results. Given what I know about polling, if the Post poll had used “some” instead of “many,” the number could easily have been 10 percentage points higher. In the future, if you plan to write a story about a poll, I recommend taking the time to read it first.