Sources of the New Deal Realignment: the Contributions of Conversion and Mobilization To Partisan Change

Abstract

T _HE SOURCES of partisan realignments have been the subject of considerable research. How do realignments occur? More specifically, who in the electorate is realigned? Two competing hypotheses have been offered in answer to this question the mobilization hypothesis and the conversion hypothesis. The mobilization hypothesis claims that the shift in the relative strengths of the parties is a result of new voters entering the electorate. People are mobilized to participate who had been too young to vote, were ineligible to vote (e.g., women prior to the Nineteenth Amendment, immigrants prior to gaining their citizenship), or were convinced that previous electoral choices were irrelevant. These new voters, the mobilization hypothesis contends, contribute more to the electoral gains of the new majority party than do the established, pre-realignment cohort of voters. Evidence supporting the mobilization hypothesis in the New Deal realignment has been found by Key (1955), Campbell et al. (1960), Sellers (1965), Converse (1976), Petrocik (1981), and Wanat (1979). The most thorough and extensive work in support of the mobilization hypothesis has been conducted by Andersen (1979a and 1979b). Although Andersen's analysis drew upon a variety of data, the core of her analysis rested on SRC party identification recall data collected from 1952 to 1972. These data were used to demonstrate that voters entering the electorate in the realignment era were overwhelmingly Democratic. Erikson and Tedin (1981) dispute Andersen's findings. They note that the partisanship recall question used by Andersen has proven to be highly unreliable (Niemi, Katz, and Newman 1980; and Reiter 1980). This problem is apparent in the distribution and trends of party identification constructed from the recall measure. According to these data, the Democrats had a plurality of the electorate as early as 1920; they gained just 3 percent of the electorate from 1924 to 1930 (Andersen 1979a: 61); and they held nearly a two-to-one advantage over the Republicans among young voters in 1924 prior to the realignment. Such counterintuitive findings raise serious doubts about the reconstructed partisanship variable. Moreover, despite her earlier arguments asserting the validity of the measure, Andersen seems to acknowledge a problem when she admits that many young Democrats of the 1920s, identified as such by the recall measure, would have identified themselves at the time as independents or without a party (Andersen 1979: 67).1 Erikson and Tedin (1981: 952) also argue that