Skip to main content icon/video/no-internet

Response Bias

The response to a survey question is said to be biased if it does not reveal, on average, the true value of the target variable. This definition is closely related to the definition of a bias in statistics: The bias of an estimator is the difference between its expected value (when computed for a given sample) and the true value of the corresponding parameter (in the population). Biased responses may arise for many types of target variables: attitudes, behaviors, preferences, and expectations; sociodemographic characteristics such as age, level of education, or labor-market status; frequencies such as the number of purchases of a good or the number of doctor visits in a specified period; and monetary quantities such as income, financial assets, or consumption expenditure. This entry examines sources and implications of response bias and presents strategies for dealing with biased responses.

Sources of Response Bias

For several decades, psychologists, sociologists, and survey researchers have worked to understand the cognitive and communicative processes that generate survey responses. A central insight is that answering a survey question is a complicated process consisting of several distinct tasks (even though these tasks are not completely sequential or independent processes). Biased responses may arise at any of these stages. It is important to note that there is no single process that would be responsible for response bias. Potential sources of biased responses can be illustrated with a conceptual model of the survey response process that distinguishes four stages.

  • First, the respondent needs to comprehend the question: He or she needs to understand what information is sought. Poor wording may easily lead to systematic misunderstanding of the question and thus to biased responses.
  • Second, the respondent must retrieve the relevant information from his or her memory. This stage activates retrieval strategies that often require filling in missing details. The success of these retrieval strategies depends on many variables, such as the respondent's cognitive ability and memory capacity.
  • The third stage involves making a judgment: Is the information just retrieved from memory complete? If not, the respondent uses estimation strategies that integrate the retrieved information with other salient information that is contained in the survey questionnaire, provided by the interviewer, and so on. These estimation strategies often involve a notion of satisficing, that is, the respondent uses just as much time and effort as is needed to construct a response that is “good enough.” Estimation strategies and the corresponding heuristics typically result in biased responses rather than random errors.
  • The final stage is to report the response that has been constructed in the first three stages. The respondent may alter his or her response because he or she does not want to reveal the constructed response for reasons of confidentiality or social desirability.

Examples

Measurement of Attitudes and Behaviors

Many survey measurements refer to attitudes or behaviors that may or may not be socially desirable. For instance, in a country in which the death penalty is considered immoral, respondents who are in favor of it may nevertheless report that they oppose it. Other examples are questions on substance use or on donations to charities. Depending on the circumstances, social desirability will result, on average, in under- or overreporting of attitudes or behaviors. Such effects are likely to be stronger in personal interviews than in self-administered mail or Internet surveys.

...

  • 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