Sunday, April 11, 2004

Concentrating Minority Voters Builds Liberal Strength in the South

STANFORD GRADUATE SCHOOL OF BUSINESS—Did liberals shoot themselves in the foot when they amended the Voting Rights Act to promote the creation of majority-minority districts? Political scientists have argued for more than a decade that creating safe liberal seats by culling large numbers of minority voters from conservative districts has inadvertently made southern congressional delegations more conservative than ever.

Are unpopular incumbent presidents likely to pander to wrong-headed popular opinions in the hope of winning re-election? Here again, the conventional wisdom says yes.

But work by Kenneth Shotts, associate professor of political economy at the Stanford Graduate School of Business, indicates that the conventional wisdom is wrong in both cases.

"The fraction of southern representatives to the left … increased after racial redistricting in the 1990s, a pattern that contrasts starkly with the well-known fact that the number of southern Democrats decreased during this period. My finding implies that racial redistricting promotes liberal policy outcomes," Shotts wrote in a paper published in The Journal of Politics in February 2003.And when he and Brandice Canes-Wrone of Northwestern University looked at presidential responsiveness, they found the expected—presidents are more responsive to public opinion when elections are imminent—and the unexpected. "Presidents with approval ratings that are significantly above or below average have the greatest propensity to take unpopular positions," they wrote in a paper that will be published in the American Journal of Political Science in October 2004.

Beyond its scholarly significance, Shotts' carefully nuanced work is important for the light it sheds on the mechanics of decision making in the White House and the House of Representatives, and for developing a definition of political "pandering" that should prove useful during what promises to be one of the dirtiest political campaigns in recent history.

To most Americans, gerrymandering is probably something of a dirty word. It implies the creation of oddly shaped congressional districts designed to give one political party an unfair advantage over another while creating legislatures that are not reflective of the voting population.

Although that's often been the case, the racial gerrymandering following the 1990 Census was designed to make Congress more reflective of the voting population by increasing the number of minority representatives.

The courts created new majority-minority districts in Alabama, Florida, Georgia, Louisiana, Maryland, North Carolina, South Carolina, Texas, and Virginia, and strengthened minority control of an existing majority-minority district in Mississippi.

But in 1994, the Gingrich revolution put Republicans in control of both houses of Congress and put the Speaker's conservative Contract with America on the front burner. Many scholars and political pundits concluded that overall, the redistricting helped Gingrich by making majority-controlled districts even more conservative than they were.

The controversial redistricting effort and its apparent misfire quickly attracted academic attention. One of those studying the topic was Kenneth Shotts, then a Stanford graduate student. "I had a theory that racial redistricting was actually pushing policy outcomes to the left. Frankly, I assumed that when I examined empirical data on the subject, I'd prove myself wrong," he said in an interview with Stanford Business.

He certainly seemed wrong. Given that Democrats suffered dramatic losses in southern House delegations in the 1990s, it would be natural to expect a similar decrease in the number of liberal southern representatives, Shotts wrote.

In fact, the opposite is true.

Shotts had embraced a theory of political decision making called "the median legislator model." Simply put, the model ranks members of Congress (or other bodies) from left to right by comparing their votes to scorecards compiled by groups like Americans for Democratic Action.

Members close to the median typically become key swing votes when the rest of the legislature is pulling in opposite directions. The median, of course, is not a fixed point. As the composition of the House changes, its median—and hence policy outcomes—move to the left or to the right.

Other theorists used statistical models to predict the preferences of representatives who would be elected under different redistricting models. Shotts instead focused on actual electoral outcomes before and after racial redistricting.

Examined in this light, it appears that the liberal strength in the southern House delegations increased noticeably between 1992 and 1996. The fraction of liberals in the South grew from an average of 38 percent in 1986-1990 to an average of 44 percent in 1992-1996, Shotts found. "I was stunned by the data," he said.

Liberals who first won election to the House in 1992 after racial redistricting took effect included Eva Clayton (NC), Melvin Watt (NC), Cynthia McKinney (GA), Sanford Bishop (GA), James Clyburn (SC), Eddie Bernice Johnson (TX), Bobby Scott (VA), Earl Hilliard (AL), Corrine Brown (FL), Carrie Meek (FL), and Alcee Hastings (FL).

Shotts explains the shift in the southern delegations this way: "From the perspective of a median legislator model of policy choice, replacing a conservative Democrat with a highly conservative Republican does not affect policy outcomes. Replacing a conservative Democrat with a liberal minority representative shifts the median, and hence policy outcomes, to the left."
In liberal states, however, the logic is reversed. Since states like New York or California are predisposed to send liberals to Washington, racial redistricting that results in the creation of some lily-white conservative districts adds representatives who fall to the right of the median, thus moving national policy outcomes to the right.

And in any case, the shift of the nation to the right during that period was influenced more by Republican victories in the Rocky Mountain states and the Senate, where racial redistricting did not occur.

In another related piece of research, Shotts and Canes-Wrone asked: when are presidents the most responsive to public opinion? When re-election time looms, of course. Shotts and Canes-Wrone knew that when they began their study of presidential responsiveness. But neither researcher expected to find that both well above-average popularity and well below-average popularity make sitting presidents less likely to take citizen opinion into account.
The researchers found this out when they examined 235 budgetary issues that were acted upon between 1972 and 1999 and for which public opinion polling data are available.

Overall, the study showed that the president and the voters agreed only 51 percent of the time, but for three of the issues—health, crime, and Social Security—they agreed more than 90 percent of the time. Agreement is relatively rare on foreign policy issues; the White House and the public agreed 32 percent of the time about defense spending and foreign aid.

This variation suggests that presidents are more likely to take popular positions on the issues that voters are more likely to encounter in their daily lives. Shotts and other researchers call them "doorstep issues."

Drilling deeper, the researchers found that:

When the next election is distant, the likelihood that the president chooses a popular policy is unrelated to his public approval. When the next election is soon and the president's popularity is below average (an approval rating below 50 percent), the likelihood of the president choosing a popular policy increases as the president's approval increases. An unpopular president can only win re-election by achieving a major policy success. He therefore has an electoral incentive to choose the policy he believes his correct, even if it is unpopular. When the next election is soon and the president's popularity is above average (60 percent or greater), the likelihood of the president choosing a popular policy decreases as the president's approval increases. Choosing an unpopular policy hurts his popularity a bit, but not enough to cost him the election, Shotts explains. Therefore, the president is likely to choose the option he believes is likely to produce a good outcome, even if voters prefer a different policy. Presidents of average popularity tend to be the most responsive to public opinion when an election looms, "even if he believes it is not in their [the voters'] best interest," and that, says Shotts, is pandering.

Leadership, on the other hand, "is doing what is in [voters'] interest, even if the people disagree," said Shotts. "The bad news is that the electoral system provides an incentive to pander; the good news is that it doesn't happen all of the time."

Other related areas that could be the subject of future research include the effect of snap elections—such as occur in parliamentary systems—on popularity, and the effect of limiting or changing the length of a president's term in office.

Related Information

The Conditional Nature of Presidential Responsiveness to Public Opinion
Kenneth Shotts, Brandice Canes-Wrone
American Journal of Political Science, October 2004

Does Racial Redistricting Cause Conservative policy Outcomes? Policy Preferences of Southern Representatives in the 1980s and 1990s
Kenneth Shotts
Journal of Politics, Vol. 65, No. 1, February 2003