This item is by Alan Abramowitz, who is Alben W. Barkley Professor of Political Science at Emory University and a member of the TDS advisory board.
As Ed Kilgore recently discussed at FiveThirtyEight.com, it has become almost an article of faith in Republican circles that the best way for the GOP to regain the ground it has lost in the last two elections is to nominate candidates who take consistently conservative positions on the issues facing the country. According to the “move right and win” theory, by standing forthrightly for traditional family values, smaller government, and less regulation of business, Republican candidates can energize their party’s base and win back conservative voters who became disillusioned with the free-spending ways of the Bush Administration and congressional Republicans.
But while the move right and win theory is extremely popular among Republican activists, it directly challenges the widely accepted view of American voting behavior among election scholars. According to the median voter theory first proposed by Anthony Downs in his seminal work, An Economic Theory of Democracy, general election candidates in the U.S. who take strongly conservative or strongly liberal positions tend to alienate moderate voters and therefore perform more poorly at the polls than candidates who hew more closely to the center of the ideological spectrum.
Fortunately, there is some readily available evidence that allows us to test these two competing theories. We can compare the performance of moderate and conservative Republican incumbents in recent U.S. Senate elections. If the move right and win theory is correct, we should find that conservative incumbents did better than expected based on the normal vote for their party while moderate incumbents did worse than expected; if the median voter theory is correct, however, we should find that moderate incumbents did better than expected based on the normal vote for their party while conservative incumbents did worse than expected.
In order to determine whether Republican incumbents did better or worse than expected based on the normal vote for their party, I measured their vote share compared with that of the current or most recent Republican presidential candidate in their state. I measured the conservatism of Republican senators based on their voting records in previous two years using a modified version of the familiar DW-NOMINATE scale with a range from 0 (moderate) to 8 (very conservative).
Figure 1 displays the relationship between the conservatism of Republican senators and their electoral performance. A positive electoral performance score means that a senator ran ahead of the Republican presidential candidate while a negative score means that a senator ran behind the Republican presidential candidate.

The results in Figure 1 show that there was a fairly strong negative relationship between conservatism and electoral performance. The more conservative the voting record, the worse the performance of the incumbent. Republican senators with moderate voting records like Olympia Snowe, Susan Collins, and John Chafee generally ran well ahead of the Republican presidential candidate in their state while those with very conservative voting records like John Ashcroft, James Inhofe and Jim Bunning frequently ran behind the Republican presidential candidate.
The results in Figure 1 would appear to support the median voter theory and undermine the move right and win theory. Before reaching this conclusion, however, we need to control for a variety of other factors that influence the outcomes of Senate elections involving incumbents such as the strength of the challenger, the national political climate, and the presence of any major scandals or controversies involving the incumbent. Therefore Table 1 presents the results of a multiple regression analysis of Senate election outcomes including all of these predictors along with the conservatism of the incumbent’s voting record. Challenger strength is measured by the natural logarithm of challenger spending in thousands of collars, the national political climate is measured by dummy variables for each election year, and scandals or major controversies are measured by a dummy variable. Once again, the dependent variable is the performance of the incumbent compared with the Republican presidential candidate.


The results in Table 1 provide additional support for the median voter theory. After controlling for challenger strength, the presence of major scandals or controversies, and the national political climate, the conservatism of the incumbent’s voting record continues to have a strong negative influence on incumbent electoral performance. For every additional one point increase in conservatism, Republican incumbents lost an additional three percentage points in support relative to their party’s presidential candidate.
Conclusions:
Evidence from U.S. Senate elections since 2000 provides strong support for the median voter theory of U.S. elections. This evidence shows that conservatism had a significant negative effect on the electoral performance of Republican incumbents. Based on these results, efforts by the Tea Party movement and other conservative activist to purge moderate incumbents and pressure Republican candidates into taking more consistently conservative positions are likely to have a detrimental impact on the GOP’s performance in future elections.