Nondifferentiation

Survey respondents are routinely asked to answer batteries of questions employing the same response scale. For example, in an effort to understand consumer preferences, respondents might be asked to rate several products on a scale of 1 to 5, with 1 being “very poor” to 5 being “very good.” Nondifferentiation (sometimes called “straight-lining”) occurs when respondents fail to differentiate between the items with their answers by giving identical (or nearly identical) responses to all items using the same response scale. That is, some respondents might give a rating of 2 to all products, producing nondifferentiated answers.

In the survey literature, nondifferentiation is identified as a very strong form of satisficing. According to the notion of satisficing, when respondents are unable to or unwilling to carefully go through all the cognitive steps required in answering survey questions, they may satisfice by looking for an easy strategy or cues to provide a satisfactory (but not optimal) answer. Nondifferentiation is such an easy response strategy that it saves cognitive effort; respondents presumably do not retrieve information from memory and do not integrate retrieved information into a judgment (or estimation). Instead, they may interpret each question within a battery superficially and select a reasonable point on the response scale and stick with that point for all items in the battery. The answers are thus selected without referring to any internal psychological cues relevant to the specifie attitude, belief, or event of interest.

Like other satisficing behaviors, nondifferentiation is most likely to occur when (a) respondents do not have the ability to answer optimally, (b) respondents are not motivated to answer carefully, and/or (c) the questions are difficult to answer. Studies have demonstrated empirically that nondifferentiation is more common among respondents with lower levels of cognitive capacity (such as respondents with less education or with less verbal ability) and more prevalent toward the end of a questionnaire. In addition, non-differentiation is more prevalent among respondents for whom the question's topic is more personally important.

Nondifferentiation may occur regardless of the mode of data collection. However, there is evidence suggesting that nondifferentiation is more likely to occur with modes that do not promote respondent motivation or use more difficult response tasks. For instance, Web surveys have been shown to promote nondifferentiating responses, especially when questions are displayed in a grid format (i.e. a tabular format where question stems are displayed in the left-most column and response options are shown along the top row). In addition, Web surveys appear to lead to more nondifferentiation than interviewer-administered modes. Within interviewer-administered modes, respondents are found to give more nondifferentiating responses to the telephone surveys than to the face-to-face interviews.

Nondifferentiation is a form of measurement error and thus decreases data quality (both validity and reliability). Of considerable concern, the presence of nondifferentiating responses artificially inflates inter-correlations among the items within the battery and thus suppresses true differences between the items. Therefore, measures should be taken to reduce the extent of nondifferentiation in a survey. Survey researchers, for example, should take measures to help increase respondent motivation to provide thoughtful answers (e.g. interviewers instructing or encouraging respondents to think carefully before answering a survey question) or to lessen the task difficulty (e.g. avoiding a grid format in a Web survey and avoid placing a battery of similar items toward the end of [Page 521]a survey) in order to reduce the extent of nondifferen-tiation in a survey.

...