Temporal Changes
While we find only little overall average differences between most demographic groups, these may mask changes in public support where some groups are becoming more supportive of Israel over time. For the most part, the descriptive trends mentioned do not support this conclusion. Religious, racial, generational and other small differences have been relatively consistent and have tended to vary in parallel (Page & Shapiro, 1992). But we require an empirical test of temporal changes that is more rigourous than descriptive plots.
To test the temporal changes statistically, we estimate our original models on each survey separately and save the coefficient (along with 95% confidence intervals) to examine if and how the coefficient changes over time. We do this only for the sympathy measure, because it is the only one that provides a sufficiently long time period with frequent measurements (favorability is also a frequently measured item, but we find no temporal changes in it for demographic groups).
We present here only three cases of interest where we find some minimal temporal changes: Race, education and religion. Consistent with most descriptive trends, we find no differences for all other demographics. For the most part, within-group change over time is rare. In most cases, coefficients remain relatively the same, and movement is minimal.
Race
Figure 5.17 illustrates this trend for race. The dots represent the effect (coefficient) from the logistic model of African-Americans and of Hispanics (whites serve as the reference category). The solid line is a Lowess smoothing line, which better illustrate the underlying trend of the coetficeint. The dashed lines represent the 95 percent confidence intervals.
The left panel plots the coeetficients for African-Americans (vs. whites). Symapthies with Israel among African-Americans have usually been lower than that of the white population. The coefficient becomes smaller over time, suggesting the large and relatively consistent gap between African-Americans and white Americans in the 1970s—1980s has, in the early 2000s, narrowed, becoming less significant over time.
The right panel plots the coefficients for Hispanic (vs. whites). Here, the gap is smaller and shows minimal change over time. Most surveys, in fact.
FIGURE 5.17 The Changing Effect of Race, Over Time
Note. Dependent variable: Sympathies in the Arab-Israeli conflict.
Dots represent the coefficients in a single survey for African-Americans (left) and Hispanics (right) compared to the white population. Solid line represents a Lowess smoothing line with a bandwidth of 0.8. Dashed lines represent upper and lower bound of the 9S percent Cl associated with each coefficient.
indicate no significant effect between Hispanics and whites, and the negative coefficient we witnessed in previous models is a result of a few surveys, especially in the early 2000s, which are negative and significantly different from zero.
Education
The pattern for education is particularly interesting. Recall that the descriptive trends on sympathies suggest that the two education groups—with or without academic education—flip in their sympathies toward Israel. Naturally, the average predicted probabilities indicate no significant difference between the two groups. The change in coefficient for educated respondents, illustrated in Figure 5.18, reveal the changing effect of education. Consistent with Gilboa (1987), educated Americans tended to symapthize more with Israel than with Arabs. Gradually, over time, educated respondents shifted, and they have become less likely to sympathize with Israel, especially in the last two decades.
Religion
Perhaps the most important trend is in the change in coefficient for evangelical Protestants. Figure 5.19 illustrates the difference between evangelical Protestants and mainline Protestants over time. The difference between evangelicals and mainline Protestants seems to grow gradually over time. Thus, evangelical
FIGURE 5.18 The Changing Effect of Education, Over Time
Note: Dependent variable: Sympathies in the Arab-Israeli conflict.
Dots represent the coefficients in a single survey for college-educated respondents compared to respondents with a high school education or less. Solid line represents a Lowess smoothing line with a bandwidth of 0.8. Dashed lines represent upper and lower bound of the 95 percent Cl associated with each coefficient.
FIGURE 5.19 The Changing Effect of Evangelical Protestants, Over Time
Note: Dependent variable: Sympathies in the Arab-Israeli conflict.
Dots represent the coefficients in a single survey for evangelical respondents compared to mainline Protestants. Solid line represents a Lowess smoothing line with a bandwidth of 0.8. Dashed lines represent upper and lower bound of the 95 percent Cl associated with each coefficient.
Christians are increasingly becoming more supportive of Israel compared to mainline Protestants. This is the only demographic divide we find that is growing over time. The results are consistent with the rich scholarly work on the emerging role of the Christian Right in support for Israel (Clark, 2007; Goldman, 2018; Spector. 2009).
