American party identification.
As with earlier work, some of these factors can be expected to
mirror those that apply in the general population.
Ample research has established the basic demography of partisanship in the United States
(see, for example, Berleson, Lazarsfeld, and McPhee 1954; Kamieniecki 1985; Knoke 1976;
Ladd 1981; Lazarsfeld, Berelson and Gaudet, 1948; Miller and Shanks 1996; Nie, Verba, and
Petrocik 1979; Shafer and Claggett 1995). Democrats tend to have lower incomes and less
education than Republican identifiers. They also are more likely to be Catholic,
members of
distinctive racial and ethnic groups and in lower status occupations. Gender differences are
minimal, with women being slightly more Democratic than men are. The effect of age on
partisanship is unclear, with the effects of life cycle and generational experience confounding
age's relationship with party affiliation.
The literature offers several processes to account for how these partisan affiliations arise.
The classic American Voter analysis (Campbell et al. 1960) emphasized parental socialization or
other early sources of affective attachment and the concomitant resistance of partisanship to
change, including that caused by policy or other short-term forces.
Parental socialization to a
party was documented by, among others, Jennings and Niemi (1968) and, more recently, Cassel
(1993). Subsequent “revisionists” argued that party was less stable than the classic picture
implied; for example, Jackson (1975) and Page and Jones (1979) found that short-term forces
(issue and candidate evaluations) could alter partisan preference.
3
It will be difficult with our data to distinguish some of the direct effects upon the vote from indirect effects via
partisanship. We have limited numbers of cases, especially in the second generation, which limits the independent
variables we can use. We also need to measure indirectly, or not at all, some key variables (e.g. ideology). We thus
chose not to estimate a full recursive model.
4
In the 1988 American National Election Study data, 52% of Protestants are Republican in contrast to 26% of Jews
and 40% each of Catholics and those with “other” religious leanings. Half of the Catholics are Democrats, versus
38% of the Protestants, 55% of the Jews, and 39% of the others.
5
Of course here, as in so many places, the argument was far more subtle and nuanced in the original than in its
retelling; chapter 10 (Campbell et al. 1960) is titled “The Formation of Issue Concepts and Partisan Change” and
includes the claim that those who conceptualize at lower levels are most susceptible to change.
4
Convention