How can 78 percent and 95 percent both be right?

An article we posted Jan. 5, “Survey finds most people support some restrictions on abortion,” detailed results of an online survey about people’s opinions on various types of restrictions and regulations on abortion.

Those who read the press release on the survey issued by the U.S. Conference of Catholic Bishops might have noticed that it highlighted different statistics than the CNS article used.

The CNS story says, for instance, that 78 percent of those who participated in the survey favor requiring abortions to be performed only by licensed physicians. The USCCB’s press release cites that “95 percent favor laws ensuring that abortions be performed only by licensed physicians.”

The data reported in each was correct and was drawn from the same survey, conducted by Harris Interactive.

Say, what?

The difference lies in two phrases.

The USCCB press release prefaced the statistics it included with the phrase “among those expressing support or opposition to the six kinds of laws examined in the survey.” In other words, of those who indicated their opinion about this law, 95 percent expressed support for it.

The CNS story included the sentence,  “Between 5 percent and 9 percent of the participants in the survey declined to answer some of the questions.” In other words, 78 percent of 2,341 participants said they support these laws, but not everyone answered the question one way or another.

Primarily because it was simpler to be consistent, we opted to use the percentages of the whole group that participated in the survey rather than the different base statistics for who answered each question, as the USCCB press release did.

In trying to report the survey results in a way that was easy to follow, it seemed simplest to stick with one number of participants across the board. The alternative would have been to include for each number cited an explanation about how many people chose to answer that question.

This entry was posted in CNS, politics. Bookmark the permalink.