Attitude Measurement

Question Format

People hold attitudes toward particular things, or attitude objects. In question format, an attitude object is presented as the stimulus in an attitude question, and respondents are asked to respond to this stimulus. Consider the following question:

The attitude object in this question is the president's handling of his job. The respondents must consider what they know about how the president is handling his job and decide whether they approve, disapprove, or neither approve nor disapprove. Another possible closed-ended format is to turn the question into a statement, and ask the respondents whether they agree or disagree with a declarative statement, for example, The president is doing a good job. However, some research indicates that the agree-disagree format produces “acquiescence bias” or the tendency to agree with a statement regardless of its content. Yet another closed-ended format is to ask the respondents to place themselves on a continuum on which the endpoints are labeled. For example, one could ask, How do you feel the president is handling his job? and ask the respondents to place their opinions on a scale, from 0 being poor to 10 being excellent.

Researchers measuring attitudes must decide how many scale points to use and how to label them. Five to seven scale points are sufficient for most attitude measures. Assigning adjectives to scale points helps define their meaning, and it is best if these adjectives are evenly spaced across the continuum.

Sometimes a researcher wants to understand the preferences of respondents in more depth than a single closed-ended question will allow. One approach for this purpose is to ask the question in an open-ended format such as, If the Democratic Party were a person, what traits would you use to describe it? Here, the Democratic Party is the attitude object or stimulus. An advantage of the open format is that the answers are not limited to the researchers' own categories. The answers to such a question will provide insights into whether or not the respondent holds positive, negative, or conflicted attitudes toward the attitude object (the Democratic Party, in this example). However, open-ended responses can be very time [Page 38]consuming to code and analyze. Alternatively, one can list a series of attributes and ask the respondent to rank them. This is easier to analyze but can be cognitively complex if respondents are asked to rank too many items.

Two other important considerations for the response options are whether or not to include a “No opinion” option and/or a middle option. Research suggests that more respondents will use both of these options when they are explicitly offered than when it is left up to respondents to volunteer such responses on their own. Research has also shown that many respondents are willing to offer opinions on obscure or fictitious issues, especially when a “no opinion” option is not offered as an explicit response choice. However, other research suggests that an explicit “no opinion” option may encourage individuals who do have attitudes to not report them. In some measurement contexts, using a middle response choice that conveys a position of noncommitment toward the attitude object makes sense. However, those who have less intense feelings or views about an issue are disproportionately influenced by the inclusion of a middle option. For this reason, the middle option is sometimes omitted, and attitude strength instead is measured with a separate question.

...