Self-Reported Measure

Self-reported measures are measures in which respondents are asked to report directly on their own behaviors, beliefs, attitudes, or intentions. For example, many common measures of attitudes such as Thur-stone scales, Likert scales, and semantic differentials are self-report. Similarly, other constructs of interest to survey researchers, such as behavioral intentions, beliefs, and retrospective reports of behaviors, are often measured via self-reports.

Self-reported measures can be contrasted to other types of measures that do not rely on respondents' reports. For example, behavioral measures involve observing respondents' behaviors, sometimes in a constrained or controlled environment. Similarly, physiological measures like galvanic skin response, pupillary response, and subtle movements of facial muscles rely on biological responses rather than self-report. Measures of other variables, such as weight, height, or cholesterol level could also be assessed without self-report by weighing or measuring respondents or by taking [Page 805]specimens like blood or urine samples, as often is done in health surveys. Finally, implicit measures such as the Implicit Association Test (IAT) and Russell H. Fazio's priming paradigm involve tasks that are not under conscious control and that do not make respondents overtly aware that their attitudes are being measured.

Historically, surveys have almost exclusively made use of self-report measures, but technological advances have made other types of measures more plausible. For example, response latencies can be measured in survey questions either by having interviewers record the length of time respondents take to answer a question. They also can be measured in computer-assisted self-administered and Internet surveys. Biological specimens or measurements can be taken in in-person interviews. Implicit measures like the IAT or priming tasks can be implemented in computer-assisted self-interviewing surveys or Internet surveys.

Although self-report measures are widely used, survey researchers using these measures should be aware that their use is based on the assumptions that respondents are able to answer the questions posed to them and that they are willing to do so, and that these assumptions may not be true. For example, people have limited and often imperfect access to many of their own internal mental processes, and they may therefore not be able to give accurate responses to questions about these processes. However, when asked about internal mental processes, respondents may construct a logical response based on their theories about their mental processes, rather than on actual knowledge of these processes. Thus, respondents will answer questions about these processes, but those answers may not be accurate reflections of the processes themselves. For example, respondents in a survey may willingly answer questions about why they voted for one presidential candidate over another, but those answers may not reflect their actual decision processes. Respondents' self-reports may also be inaccurate because they can be influenced by context, which is demonstrated by research exploring the effect of question order on survey responses. Furthermore, respondents' answers to survey questions may also be inaccurate because of limits to memory or errors in memory.

The use of self-report measures also assumes that respondents are willing to answer researchers' questions. Because being viewed favorably by others is more likely to bring rewards and minimize punishments than being viewed unfavorably, people may sometimes be motivated to construct favorable images of themselves for other people (e.g. for interviewers), sometimes via deceit. Such systematic and intentional misrepresentation by respondents when answering questionnaires has been well documented. For example, people are more willing to report socially embarrassing attitudes, beliefs, and behaviors when their reports are anonymous and when respondents believe researchers have other access to information about the truth of their thoughts and actions. Thus, some people sometimes distort their answers to questions in order to present themselves as having more socially desirable attitudes, beliefs, or behavioral histories, and people's reports may therefore be distorted by social desirability bias.

...