Voter Turnout Analysis

One of the things that stand out across all the models, is the lack of significant findings regarding state level fixed effects. This is contrary to my expectations given the differences in subregional culture across states. One of the most often mentioned subregional divides in the South, is that between the Peripheral South and Deep South. Previous research has confirmed strong differences between these two regions, especially in voting and partisanship. For example, extant research suggests that are many differences between Virginia, a Peripheral South state, and the states of the Deep South. All of this said, the findings for Tennessee are interesting. While not significant at traditional levels of statistical significance, my Tennessee variable was significant at the 90 percent confidence level in Models 5 and 6. There has been previous discussion among scholars about what states constitute the “South.” While the 11 former Confederate states are typically used, some analysists base their data on the Census Bureau’s “regions.” In that data, Tennessee is not lumped in with the other southern states. These findings are far from definitive, but they suggest that Tennessee has a stronger Republican Party, at least when looking at national elections, than do the remainder of the southern states. If these findings are supported by further analyses, scholars may need to drop Tennessee from analyses of voting in the South.
Turnout Effects
Next I turn to the effects of voter turnout.
First, Models 1-4 suggest that turnout does not have a significant impact on party strength. In the national party strength models, turnout does indeed have a significant and positive impact on party strength. This finding makes perfect sense. At the start of the period of study, whether one voted was highly dependent on class. Today, this relationship is not as strong. Figures 3.3 and 3.4 show a distinctive upward trend in the data. Moreover, midterm effects are visible. It makes it reasonable to expect that more new voters were showing up at the polls and voting for at least some Republican candidates in high profile races. Turnout is not by itself significant in any of the state party strength models. This suggests that candidates such as Senator Trent Lott and President Ronald Reagan may have been able to convince new southern voters to show up and vote for them, but those new voters continued to vote for Democratic candidates in state level
races. It is widely known that the electorate in midterm elections is older and whiter than it is in presidential year elections.
To control for these effects on turnout, I included the variable midterm in all models, as well as the interaction term midterm* percentage turnout in Congressional elections. The data do not show any statistical relationship between midterm and midterm* percentage turnout in Congressional elections in the party strength or national party strength models. Interestingly, both variables are significant in the state party strength models. Midterm is negatively signed and statistically and substantively significant. During midterm elections, party strength for elected state officials was depressed 0.103 (on a scale out of 1, this is a sizable disadvantage to the Republican Party). Running counter to this finding, however, is the fact that the interaction term midterm* percentage turnout in Congressional elections was positive and significant. To put this in perspective, this means that an increase of 20 percentage points in turnout during a midterm election would increase state party strength by 0.067 out of 1. Increases in the size of midterm turnout during the period under study were seen in virtually all of the southern states, and this increase helped the Republican Party. In all, this means that the assumption that increased turnout in the South benefits the Democratic Party is false. These findings indicate that instead, that increased turnout
benefited the Republican Party, but only in midterm elections.
The Effects of Race
I included two different measures of race in each model: black registration and percentage non-white population. I find that both black registration and percentage non-white population are positive and statistically significant in the state party strength models. This supports the hypotheses of both racial threat studies and relative advantage studies. However, until the most recent decade, the nonwhite population during the time under study went down. This suggests that the exodus of black people to the Northeast and Midwest should have reduced state party strength. But neither of these two measures of racial composition is statistically significant in the models of party strength and national party strength. This makes the state party strength findings suspect. WHY
The Effects of Polarization
I use three different measures of polarization in my statistical models.
The coefficient on House party difference in Model 1 does not reach statistical significance. House party difference is also used in Model 4. In this model, the coefficient on this variable again does meet statistical significance. Figure 3.6 displays data showing that in the 1950s this measure was below 0.5 and had risen to above 1.0 by 2010. With a technical range from 0 to 1 for party strength, the 0.5 change in the Polarization of the House of Representatives that was seen during the period of study would account for an increase of 0.034 in party strength, which is not
insubstantial.WHY
Model 2 includes a variable measuring polarization of the Senate. The coefficient on Senate party difference does not meet traditional levels of significance in Model 2, but does predict a sizable impact on party strength. In all, Model 3 shows that party strength was not strongly affected by the stances of local members of Congress. I reach this conclusion on the basis of insignificant coefficient on Southern Democrats. I do, however, see in the models of national party strength and state party strength that all of the coefficients are in the anticipated direction. In addition, in three of the four models, the coefficient on Polarization of the House of Representatives is statistically significant.
The Effects of Religion
Turning next to the variable Percentage Southern Baptist, we see that it is not significant in any of the models. Part of the reason for the lack of significance is that the standard errors for the variable are very, very large. I had hoped that the data would show that changes in the largest conservative Christian denomination in the region would increase party strength, as there is an extensive literature on the relationship between faith (denominations, strength, and activeness) and partisanship/voting (which is a similar concept to party strength). There is no evidence of that relationship in these models. This finding does not by any means preclude religion from having an impact on party strength. Instead ,it just means that fluctuations in the largest conservative Christian denomination do not predict changes in party strength. Other alternatives that could have lead religion to impact party strength include an increased issue awareness among Christians (particularly on issues like gay marriage and abortion) resulting in conservative Christians changing party allegiance.
Conclusion
The positive findings of the first three models provide evidence that the voters from southern states adjusted their voting patterns due to changes in the economy. In this case, the change in the economy was the decline of agriculture and the increase of southerners’ incomes. This response appears to be the result of the change of the structure of the southern economy from an agricultural economy to a diverse modern economy. It can be seen in the other models that this process occurred for both elections to state and national offices
I also included in the analysis controls for alternative explanations of the partisan shift in the South. There were some interesting findings. Some of these findings could only be discovered because party strength could be bifurcated into state party strength and national party strength.why
This was not the only change that resulted in an increased ability for the Republican Party to receive votes in the South. The analysis showed that the polarization of congress and changes in turnout also affected the voting public. Several of the models showed that polarization affected party strength. Turnout had several unique results. The models showed that increased turnout benefited the Republican Party in elections for national offices, but if anything hurt them in elections to state level offices. Moreover, while the Republican Party started out with midterm elections being bad for its candidates, as turnout increased during the midterm elections this benefited the Republican Party.
In the next chapter, I will discuss problems with the analyses and how these problems may have affected the findings reported here. Of particular note, I will discuss issues that may have depressed the findings for all the models in this chapter. Nevertheless, the models did produce positive findings that supported my primary hypothesis. The implications of these findings are obviously large, but they are tempered by the lack of significance across all specifications.