“Sampling public opinion is like sampling soup: One spoonful can reflect the taste of the whole pot if the soup is well-stirred.” — George Gallup

To collect the viewpoints of the 75,000 teachers in the UFT, we conducted our survey in much the same way Gallup Inc. conducts its national polls: We used random or probability sampling.

Random sampling is similar to pulling names out of a hat in that every member of the population has the same quantifiable chance of being selected. This means that no subset of the population, such as middle school teachers or teachers with fewer than three years of experience, is favored or excluded from the sample. Rather, the sample is representative of the entire population and subgroups have the same proportional weight in the sample as they do in the population [see “Who answered the survey,” right].

Image

For this year’s survey, we sampled only from the UFT’s 66,600 general and special education teachers who work in school settings, which meant that every teacher had less than a 1 percent chance of being selected. From this random sample, we got the names of 2,510 teachers who were invited to take the survey.

We received responses from 836 teachers, for a response rate of 33 percent, substantially better than the 15 percent that is typical for online surveys. When members of a sample don’t respond, it affects the results by increasing the potential for error — the possibility of over- or underestimating the strength of a finding. We believe that the margin of error on most of our questions is no more than plus-or-minus 10 percentage points, given the nonresponse rate we experienced.