An Examination of the Measurement Adequacy of the NES Scales for the Measurement of Racially Relevant Attitudes


  • The author wishes to acknowledge Kevin Arceneaux, Christopher Wlezien, Richard Joslyn, Greg Szczepanek, and David Ford, all of Temple University, for their help at various stages of this project. Of course, ultimate responsibility for the contents of the article fall on the author's shoulders only. The data analyzed here were gathered by the ANES at the University of Michigan in their 2008 National Election Study. The author bears all responsibility for the uses and interpretations of these data.

Direct correspondence to Michael Hooper, Science Gladfelter Hall, Temple University, Philadelphia, PA 19122 〈〉.



The objective of the research reported in this article is to determine if the NES scale for the measurement of equalitarianism is adequate for the measurement of attitudes toward equality in the mass public.


The analysis begins with the conjecture that the items are framed in terms that are too abstract for the typical member of the mass public to comprehend. As a result, it is argued, each item will not measure the same latent variable, equalitarianism. Rather, item scores will be saturated with systematic error in the form of response set, with the items falling into separate groups depending on how they are framed, rendering scores on the items problematic for use in empirical analysis. For contrast, responses to the items constituting the NES racial resentment scale are also examined. The items comprising this scale are framed in more concrete terms. As a result, it is argued that items comprising the racial resentment scale should display much less systematic error in the form of response set when used in the mass public. Using NES data from a national sample drawn in 2008, responses to the items comprising the equalitarian and racial resentment scales were analyzed via differences of means, correlations, and LISREL (C)onfirmatory (F)actor (A)nalyses.


The results for the equalitarianism scale do not support the conclusion that each item is measuring the same latent variable. Rather, the results are consistent with the argument that response set is the dominant factor in shaping responses. The results for the racial resentment scale support the conclusion that each item is measuring what it is designed to measure and response set is playing a relatively minor role in shaping responses. An additional confirmatory factor analysis of scores on the equalitarian scale was done using only highly educated respondents. This segment of the population can be expected to deal relatively easily with items framed in abstract terms. The CFA for this group of respondents yielded results that bear out this expectation and are interpreted as buttressing the principal argument of the article that the items constituting the scale are not suitable for use in the mass public because they are couched in terms too abstract for the typical member of the mass public to comprehend.


The article concludes with the suggestion that it would be useful to develop a new scale for the measurement of equalitarian attitudes in the mass public, along with some suggestions regarding the nature of such an instrument.