Partisanship over Time, 1970-2004

duped into voting Republican due to social issues, his so-called “working class conservative.” While there are significant polarization trends for the other groups and for income overall, substantively

There has been no substantial uptick in class polarization on partisanship, as is evident in Figure 10.1. The Class GP on partisanship indicates just a slight increase in polarization on party ID. Overall, there is only a 0.016 increase in polarization for every survey year, mostly a product of the moderate spike in class polarization in 1994 and 1996. Furthermore, the best fitted model for increasing polarization is not for the poor or working class but rather for the “rich” (R² = .325). Not only is the “rich” category the best fit, but the decline in goodness of fit for the polarization model declines almost monotonically as we

Table 10.2: Trend Regressions of Income Group Polarization on Party ID, 1970-2004

Model: GP(pd) = B0 + B1(year) + e	Trend: Polarization Y/N	Intercept (S.E.)	Parameter Estimate (S.E.)		R2	N
Poor	N	-1.822 (1.598)	0.001 (0.001)		.086	17
Working	Y	-10.855 (4.596)	0.006 (0.002)	**	.277	17
Lower Middle	Y	-9.533 (4.999)	0.005 (0.003)	*	.202	17
Upper Middle	Y	-5.500 (2.942)	0.003 (0.001)	*	.195	17
Rich	Y	-3.379 (1.278)	0.002 0.000	***	.324	17
Income Group Polarization	Y	-31.089 (0.016)	0.016 (0.006)	***	.329	17

move down the income scale. The worst fit is for the upper middle class category (unsurprising given the weak polarization scores apparent in Table 10.1). So while there is a polarization trend that meets the standards of statistical significance for four of the five categories, there is very little apparent income group polarization.

The weighted and unweighted contributions of each of the class categories to the group polarization measure on partisanship can be found in Tables 10.3 and 10.4 respectively. If one examines the weighted group polarization percent contributions for the class categories, it is apparent that there has been relatively little change in the contribution of the Poor to polarization. And, in fact, that contribution peaked early in the time series (1972). Indeed, the first two years, 1970 and 1972, are the two largest contributions of the “Poor” category to group polarization. The lows are relatively recent (1992 and 1998) and the contribution of the Poor doesn’t exceed 17% at any point after the 1972 survey year. The working class numbers are fairly stable as well, except for the unusual contribution figure in 2000, where the working class contributed 52 percent of the group polarization measure for that year. The unusual years for the lower middle class were 1974 (59.28) and 1998 (58.345), all likely strongly influenced by variation in group size. For example, the lower middle class unweighted figure for 1974 was 33.69%, more than twenty percent less than the weighted number. The “Rich” are downgraded in their percent contribution due to their small size relative to the population, as is evident in Table 10.3. No apparent trend is evident from 1970 to 2004 for the contribution of the rich to the group polarization measure on partisanship. It ranges from a low of three percent to a high of 11.47% in 1980.