23
CHAPTER 4: ATTITUDE MEASUREMENT
Chapter objectives
This chapter details the most commonly used scales for attitude measurement and presents
alternatives including unobtrusive and physiological measures. Suggestions are provided for how
to write good attitude items. Further, the pros and cons of using each method are discussed.
Terms and issues to know
Context and wording effects
Guttman scale
How to ask good questions
Implicit Association Test
Likert scale
Open-ended measure
Survey measurement of attitudes
Unobtrusive measures
Chapter outline
I. There is a science to writing attitude questions, one that calls on principles of measurement,
statistics, and cognitive psychology (Hippler, Schwarz, & Sudman, 1987; Tourangeau &
Rasinski, 1988)
A. Core assumption – that one can measure phenomena by assigning numbers to objects
on the basis of rules or guidelines (Stevens, 1950)
B. There are three standard attitude scales:
2. Guttman
3. The Semantic Differential
II. Questionnaire Measures of Attitude
A. Likert Scale a series of opinion statements that asks individuals to rate their agreement
or disagreement with each statement along a numerical scale
2. Presumes that there are equal intervals between categories
4. Most researchers prefer 5- or 7-point scales because they allow respondents to
indicate shades of gray in their opinions, but do not provide so many categories that
people feel overwhelmed by choices
B. Guttman Scale scale progresses from items easiest to accept to those most difficult to
endorse; it is not easy to administer
1. Those who get a high score on a Guttman scale agree with all items
3. Those with mildly positive attitudes agree only with items that are easy to accept
4. It is hard to construct
C. Semantic Differential created by Charles Osgood, where participants rate an object
based on bipolar adjectives:
24
2. Osgood decided that people rate concepts based on three dimensions: evaluation,
potency, and activity
III. Pitfalls in Attitude Measurement
A. No perfect attitude scale
B. Inaccuracies result from such factors as:
1. Respondent carelessness in answering the questions
3. A tendency to agree with items regardless of their content (Dawes & Smith, 1985)
C. The way the researcher words the question and designs the questionnaire can elicit from
Presser, 1981; Schwarz, 1999)
1. Manner in which the question is asked can influence the response that the
researcher receives
3. Wording –
the issue
IV. Policy Implications
A. The way a question is worded or framed can shape the contours of policy debates
B. Polls have obtained strikingly different results, depending on the way questions have
D. Can be troublesome when the goals of persuasion clash with larger interests of the
country
E. In many cases the -blown attitudes at all
1. Sometimes individuals construct their attitudes on the spot, based on what happens
2. People do construct attitudes on the spot in response to
V. Asking Good Questions
A. Use words respondents can comprehend
B. Write specific and unambiguous items
C. Avoid double negatives
D. Pretest items to make sure people understand your questions
E. Ask questions in different sequences to check out order effects
VI. Open-ended measures
A. Scales do not always shed light on the underlying dynamics of attitudes– the rich
underbelly of cognitions and emotions
B. These components can be measured through more open-ended, free-form techniques
C. One open-ended technique involves assessing cognitive responses to communications
E. These responses could be subsequently categorized by researchers according to specific
criteria (Cacioppo, Harkins, & Petty, 1981)
F. Affect:
1. Can also be assessed in an open-ended way
G. Combining open-ended measures with traditional attitude scales increases the odds that
researchers will tap attitudes accurately and completely
VII. Indirect Methods to Measure Attitudes
A. Unobtrusive Measures
2. Behavior is used as a surrogate for attitude
3. Can be useful in cases where it is not possible to administer self-report scales or one
B. Physiological Measurements
2. Galvanic skin response a change in the electrical resistance of the skin
4. Facial electromyographic techniques (EMG) movement of facial muscles
5. Response time length of time it takes to agree or disagree with a statement
a. The longer it takes to answer, the less developed the attitude
b. Implicit Association Test (IAT)/ Response Time
i. Measures the latency or length of time it takes people to indicate if they
VIII. Conclusions
A. Three venerable scales are typically employed: Likert, Guttman, and the semantic
differential
B. The Likert scale is used most frequently because it taps beliefs and can be constructed
easily
C. Open-ended measures, such as cognitive responses, can supplement closed–ended,
structured scales
D. There is a variety of problems in measuring attitudes through self-reports, including
survey context and wording effects
Discussion points/ projects/ exercises
1. What are the benefits to measuring attitudes empirically? How might empirical evidence
be used to inform policy?
2. Think of an attitude you would like to measure. Create a Likert, Guttman, and Semantic
Differential scale to measure aspects of that same attitude. Have family or friends fill out
each survey type. What were the difficulties you had in creating each? Did they measure
what you intended? How would you improve each scale next time?
4. Polls have obtained different results depending on the way questions have been framed.