Always look twice at stories reporting polling results

During our daily research for The Rural Blog, our daily digest of events, trends, issues, ideas and journalism from and about rural America, we came across a press release from the Texas teachers’ union, headlined “One-third of teachers moonlight to support families.”

My blogger wrote an item that began, “Thirty-one percent of Texas teachers have to work a second job during the school year—49 percent work during the summer—to make ends meet, says a survey by the Texas State Teachers Association.”

My B.S. detector went off. Did that many teachers really say they had to work a second job to make ends meet, or to give their family the support it needed, as the release implied?

I asked TSTA for the questionnaire used in the survey of 837 Texas teachers by a professor at Sam Houston State University. You can see the pertinent part of it here. It asked teachers if they had an extra job during the summer or the school year, how many hours it took, whether their quality of teaching would improve if their teaching salary allowed them to give up moonlighting during the school year, and how big a raise would allow them to quit moonlighting.

None of the questions said anything about teachers’ need to have a second job – whether to support their family to make ends meet, to maintain the lifestyle they thought their family deserved, or whatever. While the headline on the press release was accurate – a second job presumably supports the family – its implication led my blogger to make an unjustified leap, saying the moonlighters had to make a second job to make ends meet.

That’s what many press-release writers hope reporters will do: make a stronger point that helps the cause of the entity issuing the release. This is a lesson to avoid that – and to ask for the questionnaire on which a survey is based. It’s just good reporting. In this case, we saw that the teachers were asked whether they taught in an urban, suburban or rural district, so we asked TSTA to break down the results by those categories.

It’s especially important to get the questionnaires of election polls, which can be skewed by the sequence and phrasing of questions. You deserve to see every syllable spoken to the poll respondents, up to and including the last result for which a result is provided. And the pollster should personally certify to the accuracy of the poll and be available to answer questions about its methodology.

If you need help evaluating polls, email me at [email protected].



By Al Cross

Al Cross edited and managed weekly newspapers before spending 26 years at The (Louisville) Courier-Journal and serving as president of the Society of Professional Journalists. Since 2004 he has been director of the Institute for Rural Journalism and Community Issues, University of Kentucky. See

A print-only version of this blogpost ran originally in Publisher’s Auxiliary. It is used by permission of Pub Aux and the National Newspaper Association.