Frequently Asked Questions

Demographic and geographic quotas were set, based on data from the US Census, to ensure that the population of survey respondents matched the demographic characteristics of Missouri adults. The data were statistically weighted to adjust for any remaining differences.

Because respondents were contacted on mobile phones, it was possible to reach and incorporate the opinions of residents who are unstably housed.

Interviews were conducted on wireless phones as well as on landlines; in fact, more than four in five phone interviews were obtained on wireless phones. Nonetheless, response rates to telephone surveys have steadily fallen over the years; for that reason, the survey was also offered online to accommodate Missourians who prefer to respond in that way. Roughly half of responses were obtained from Missourians who replied to an email or text invitation and chose to fill out the survey online.

Oversampling entails intentionally conducting additional interviews with demographic or geographic subgroups that may be too small to be adequately analyzed as part of a general, random sample of the whole population. Once interviews are complete, the data is statistically weighted to reflect the true demographic and geographic profile of the overall population, but researchers can still analyze the responses of oversampled subgroups with confidence that the additional interviews yield greater precision in the understanding of their responses.

The survey employed proven techniques for sampling, question design, and analysis that are widely accepted within the industry. Surveys using these techniques have been shown (for example, by comparing their findings with election results) to be accurate representations of public opinion.

Every poll, no matter how well-designed, is subject to a wide range of types of error and bias that can impact the results. Sampling error relates to the specific set of individuals who are selected to participate in a poll, who may or may not perfectly mirror the opinions of the population from which they are drawn (see the discussion of the margin of error below). Different survey firms use different procedures or question wording that can affect the results, a source of error often referred to as “house effects.” Certain kinds of respondents may be less likely to be sampled or respond to some surveys, a factor known as “non-response bias.” Respondents might not be candid about controversial opinions when talking to an interviewer or might answer in ways that reflect how they believe they should think or act, rather than how they do (such as claiming to be registered to vote when they are not), a phenomenon known as “social desirability bias.” Finally, it is important to remember than any poll is a snapshot of a moment in time – public opinion can and does change, and it is not uncommon for public opinion to evolve after a poll is completed.