Skip to main content icon/video/no-internet

Questionnaire-Related Error

Error related to the questionnaire is one component of total survey error. This type of error is traditionally viewed as being a discrepancy between the respondent's answer and the “true” score or answer. Questionnaire-related errors may stem from the content of the questionnaire (the wording, ordering, and/or formatting of the questions themselves), the method of delivery (whether by interviewer, by telephone, on paper, by computer, or some other mode), the materials that accompany the questionnaire (such as show cards in a face-to-face survey), or all of these things. The focus in this entry is primarily on response error by respondents, although these other factors can be considered secondary aspects of questionnaire-related error.

Recent work on questionnaire error draws on cognitive psychology to inform the design of standardized instruments. One prominent framework of the survey response process (or the mental steps a person undergoes when answering survey questions), advanced by Roger Tourangeau, Lance Rips, and Ken Rasinski, consists of four major components: (1) comprehension and interpretation of the question; (2) retrieval of relevant information from memory; (3) judgment, or deciding on an answer; and (4) response, including mapping the answer to response categories and reporting the answer.

Errors may arise at any step, and the questionnaire may affect the likelihood and extent of those errors. Although there are many design principles that minimize error, often there are trade-offs that researchers need to consider when adopting any given solution.

Comprehension

Taking steps to ensure question comprehension by respondents is often the first stage in developing a questionnaire that minimizes error due to misunderstanding or misinterpreting questions. In general, the more complex or ambiguous the task, the more error there will be. Questions may contain words unfamiliar to the respondent, may have complex syntax, or may assume a higher level of respondent cognitive ability than is warranted. Sometimes researchers use specialized terminology rather than colloquial language. Respondents may also not pay close attention and may miss part of the question.

One difficulty is that many words in natural language are ambiguous, and researchers often must use vague quantifiers that have an imprecise range of application. Some examples include categories of words denoting frequency (e.g. never, sometimes, pretty often, very often), probability expressions (e.g. very unlikely, somewhat unlikely, likely), amounts (e.g. none, some, many, all), and strength (e.g. extremely dissatisfied, very dissatisfied, somewhat dissatisfied). There is evidence that respondents have varied associations with words that denote frequency, regularity, or size. For example, How often … ? results in variability based on how respondents interpret the word often. It is advisable to use more specific quantifiers, such as On how many days … ? or How many times in the last week/month/year … ?

Simple words or categories that appear to have common understandings, such as household or sibling, can be interpreted differently. When possible, such terms should be defined so as to minimize variability in how respondents interpret them. For example, household can include persons who use the address as their permanent address but who are not physically living there, such as college students who live in dormitories, individuals on military duty, or individuals who may reside at that address for only part of the year. Sibling may include biological siblings, half siblings, and adopted or stepsiblings.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading