Our article, “Party Polarization and ‘Conflict Extension’ in the American Electorate.” American Journal of Political Science 46(4):786-802, by Layman and Carsey, contains a few errors. One is a printer's error while the others are minor errors in some of the tables. We sincerely apologize for these mistakes, but we want to make it clear that these errors in no way change the substantive findings of our paper.

Below is a brief description of the errors. A complete copy of the corrected Figure and Tables is available HERE. A copy of the pageproofs of this paper is available HERE or on my WORKING PAPERS page.

The printer's error is on p. 793. The values on the y-axis on the bottom-right graph in Figure 1 should range from 0 to .5.

The tables in the article also contain some errors discovered by us when replicating our results. Again, these errors do not alter any of the substantive conclusions we reached.

In table 2, on page 792, the confirmatory factor analysis for 2000 included attitudes toward “welfare spending.” That measure has a standardized factor loading of .58 in the single-factor solution and .61 on the social welfare factor in the three-factor solution. All of the other factor loadings and goodness-of-fit statistics are correct.

In table 3, on page 795, the number of observations for weak and leaning partisans in 2000 should be 579. There are a handful of errors in the factor correlations from three-factor models for pure independents, and the factor loadings for this group in 2000 are incorrect. However, the main conclusions drawn from table 3 – that the relationship between social welfare, racial, and cultural attitudes was weaker from 1988 through 2000 for pure independents than for party identifiers, and that those relationships did not grow over time for pure independents – remain true.

In table 5, on page 798, the correlation between social welfare and cultural attitudes in a two-factor model should be .03 for strong partisans who are aware of party differences on neither the social welfare nor the cultural agendas and -.05 for strong partisans who are aware of party differences on only one of those agendas. Neither correlation is statistically significant. The factor loadings for pure independents who are aware of party differences on both agendas are incorrect. However, the average standardized loading of all issues on a single factor for that group is correct. Also, the basic point that the relationship between social welfare and cultural attitudes among pure independents does not increase with awareness of party differences, as it does among party identifiers, remains true.