Semantic Differential Scale

The semantic differential (SD) is a technique developed during the 1940s and 1950s by Charles E. Osgood to measure the meaning of language quantitatively. Words may have different meanings to different individuals as a function of their experiences in the world. For example, “poverty” has been experienced differently by 7-and 70-year-olds, and by the rich and the homeless. Their expressions of understanding of poverty are modified by these experiences. The SD captures these different meanings by providing some precision in how our understanding of words differs.

The typical semantic differential test requires a subject to assess a stimulus word in terms of a series of descriptive bipolar (e.g., good-bad) scales. The subject is asked to rate the stimulus between the extreme and opposing adjectives that define the ends of these scales. Typically, these bipolar scales have 5 or 7 points. The odd number allows the subject to choose a midpoint or neutral point (which would not be available if an even number of options were used). The various scale positions may be unlabeled, numbered, or labeled (see Table 1).

Respondents seem to prefer the labeled version.

Typically, a stimulus will be assessed in terms of a large number of these bipolar scales. For example,

The pattern of responses to these bipolar scales defines the meaning subjects attribute to the stimulus word.

Osgood's early research into the usefulness of the SD suggests that respondents' ratings of stimuli on these bipolar scales tend to be correlated. Osgood argues that the semantic space is multi-dimensional but can be represented efficiently by a limited number of orthogonal dimensions. Three dimensions typically account for much of the covariation found in responses to the individual bipolar scales. These dimensions have been labeled Evaluation, Potency, and Activity (EPA) and have been found in a wide array of diverse studies. One way to interpret this finding is that, although the meanings of the bipolar scales may vary, many are essentially equivalent and hence may be represented by a small number of dimensions. To the extent that meanings are equivalent across groups, space, or time, the similarities and differences of groups, cultures, and epochs may be examined.

In practice, the semantic differential may be used to develop a nuanced understanding of attitudes. For example, Americans often hold ambiguous opinions of their national institutions and political leaders. Yet citizens are often asked their assessment of a political institution or elected official in a single question. One common question used to examine public perceptions of the American president is, “Do you approve or disapprove of the way [name of current president] is handling his job as president?” Using the SD approach, it is possible to develop a more complex appraisal of public opinion by asking respondents to rate the president on a series of bipolar scales (see Table 2):

Responses to each scale may be examined individually, or they may be subject to data reduction (e.g., factor analysis) to explore public assessments of the president on the EPA dimensions.

...