The Singer, Not the Song

In 2010, social psychologist Eric Knowles and two colleagues published a study showing that some of the opposition to health care reform results from prejudice toward African-Americans and dislike of Barack Obama. The research was a panel study in which participants were interviewed several times over the internet.

During the first contact, Knowles measured implicit prejudice against blacks using a variation of the Implicit Association Test. This test measures an automatic tendency to associate white Americans with the concept “good” and black Americans with the concept “bad.” This bias this is unintentional and occurs without our awareness. Knowles found that the people high in implicit prejudice toward African-Americans reported more negative attitudes toward Obama before the 2008 election and were less likely to vote for him. This is one of several studies to show that racial prejudice influenced votes in both the 2008 and 2012 elections.

These negative attitudes toward blacks spilled over onto Obama's policies. People high in implicit prejudice were more opposed to health care reform in 2009, before the Affordable Care Act (ACA) was passed. How do we know their opposition to health care reform was due to prejudice rather than political conservatism, which is highly correlated with prejudice in this country? Knowles did an experiment in which he described a health care plan. For half the participants, it was presented as Bill Clinton's 1993 plan and for the other half, it was presented as Obama's plan. (The actual description was of features that both plans had in common.) Implicit prejudice had no effect on attitudes toward the “Clinton” plan, but when it was attributed to Obama, the more prejudiced participants were more opposed to it. This study was replicated two years later with the same results.

Fast forward to 2014. Aaron Chatterji and colleagues just published a study asking why members of the House of Representatives did or did not vote for the ACA. The study only included Democrats, since only one of 177 Republicans voted for the bill. The researchers looked at whether three variables were related to the legislators' votes: (1) the percentage of their constituents without health insurance, (2) Obama's margin of victory or defeat in the 2008 election in their district, and (3) political contributions from health insurance companies. The statistical analysis also controlled for eight demographic variables, such as the age and racial composition of the district, and five Congressperson characteristics, such as their own 2008 margin of victory. The results were:

• Percentage of constituents without health insurance was unrelated to the legislators' votes.
• Obama's margin of victory made a significant difference. The 219 Democrats who voted for the ACA came from districts in which Obama's average margin of victory was +30%, while the 39 Democrats who voted against it came from districts in which Obama lost by slightly under 10%. Obama's margin of victory accounted for 47% of the variance in these Representatives' votes.
• Political contributions from the health insurance industry also had no relationship to voting.

The authors note that if the Congresspeople had the best interests of their constituents in mind, there should have been greater support for the ACA from Representatives whose districts contained a higher percentage of uninsured people. At different places in the article, they refer to this as either ignoring their constituents' preferences or ignoring their constituents' needs. The latter is more accurate, since they have no measure of voter preference. Maybe some of the people who needed the ACA did not prefer it (or did not know they preferred it). It is primarily the needs of their poorer constituents that these legislators ignored. This is no surprise, since there is a growing body of research showing that politicians votes are consistent with the opinions of their constituents in the top third of the income distribution, but the opinions of the lower and middle thirds are disregarded.

Of course, Obama's margin of victory or defeat is also a salient indicator of constituent preferences, and these Congresspeople were very responsive to it. However, the presidential election was not a referendum on health care reform, which played only a minor role in the campaign, but is more reasonably regarded as a measure of Obama's popularity. It appears that the legislators voted for or against the ACA based on the evidence of Obama's popularity in their district.

The common thread among both studies is that both citizens' and legislators' attitudes toward health care reform seem to be less influenced by the substance of the policy than by attitudes toward the President himself. At the present time, politicians in 24 states (including Pennsylvania) are ignoring the needs of their poorer citizens by refusing to implement the ACA's provision to expand Medicaid. Is this decision also driven by attitudes toward the President?

By the way, the fact that political contributions from insurance companies had no effect on voting doesn't really contradict the hypothesis that politicians are influenced by campaign contributions. The health insurance industry never clearly favored or opposed the ACA, since it has both advantages and disadvantages for them. In fact, the authors never predicted whether health insurance money would make a Congressperson more or less likely to vote for the bill.

You may also be interested in reading:

Old-Fashioned Racism

Whose Opinion Matters?

Symbolic Racism and Gun Ownership

Old-Fashioned Racism

A new study published by Michael Tesler in the Journal of Politics suggests that, for the first time in many years, old-fashioned racism is having an effect on party preference and voting behavior in the United States. Specifically, old-fashioned racists are more likely to be Republicans and vote for Republican candidates.

Old-fashioned racism (OFR) refers to the endorsement of statements that are obviously and blatantly racist, such as the belief that blacks are genetically inferior to whites, that blacks and whites should be segregated and maintain social distance from one another. OFR has declined steadily in this country since World War II. The most likely reason is that OFR has always been positively related to age. Old-fashioned racists have not changed their attitudes; they have died. Another contributing factor in the decline of OFR is that it has become less socially acceptable to state obviously prejudiced attitudes in a survey.

Sometimes the mass media present the decline of OFR as evidence that prejudice is no longer a problem in this country. However, OFR has been gradually replaced by modern (or symbolic) racism (also known as racial resentment). Modern racism is prejudice revealed in subtle, indirect ways, such as the claim that blacks do not respect traditional American values, or in opposition to social policies perceived to help black people. It allows people to express beliefs and endorse policies that are harmful to African-Americans, but still deny being prejudiced. Modern racism has been related to voting behavior and party preference ever since the civil rights era. Richard Nixon's “Southern strategy” was designed to help the Republican party to take advantage of racial resentment, and the strategy remains successful today.

Tesler begins his article by noting that OFR has never been related to white Americans' partisan preferences in the post-civil rights era. Although modern racists have tended to vote Republican, old-fashioned racists have been evenly distibuted among the two parties. However, that appears to have changed since Barack Obama became president in 2008. To make this point, Tesler presents three studies.

The first study looked at the effect of OFR on candidate preference in the 2008 election. Tesler reanalyzed data collected by the Pew Research Center in early 2008. In this study, OFR was measured by endorsement of items stating that whites and blacks should not date or intermarry, an undeniably racist belief still held by about half the white population. The study asked participants their preferences in two hypothetical presidential contests, John McCain vs. Hilary Clinton, and John McCain vs. Barack Obama. Results showed that OFR had a significantly greater effect on the McCain-Obama contest than the McCain-Clinton matchup. The difference between those highest and lowest in OFR was 10% on the McCain-Clinton question, but 35% on the McCain-Obama question. This difference was statistically reliable even after controlling for the effects of party preference, political ideology and modern racism.

 All rights reserved by Huffington Post

The remaining two studies looked at the spillover effect, sometimes referred to as the racialization of politics. It refers to the possibility that racial attitudes will affect political behavior in contexts having nothing obviously to do with race. For example, since President Obama is associated with health care reform, the spillover effect predicts that old-fashioned racists will oppose the Affordable Care Act, even after controlling for the influences of party preference, political ideology and modern racism. 

In Study 2, the Pew data were used to examine the relationship between OFR and party preference from 1987 through 2009. From 1987 through 2007, there was no significant relationship between OFR and partisanship. However, in 2009, there was a significant association between OFR and self-identification as a Republican. Further analysis showed that, consistent with the spillover hypothesis, this relationship is mediated by attitudes toward President Obama. That is, when the effect of evaluation of President Obama is statistically removed, the association between OFR and partisanship disappears. (See my earlier post for a more thorough explanation of mediation.)

Finally, Tesler did a survey of his own in which respondents were asked to state their preference for candidates for the House of Representatives in the midterm election of 2010. This study contained a manipulation designed to test the spillover hypothesis. Before stating their voting preference, a randomly selected half of the respondents were reminded that President Obama had been campaigning for Democratic candidates in the midterm election and asked whether this affected their preference. This reminder, called the “Obama prime,” was omitted for the other participants. Without the Obama prime, OFR had no effect on voting intentions in the midterm election. However, with the Obama prime, those highest in OFR were 13% less likely to prefer the Democatic candidate than those lowest in OFR.

It is important to note that Tesler is not saying that OFR has increased in recent years. As far as I know, it is still slowly declining. However, prior to 2008, old-fashioned racists were randomly distributed among the two political parties. Since many of them are what are euphemistically called “low information voters,” some of them self-identified as independents. The effect of the Obama presidency and related changes in the political culture, such as the rise of the Tea Party, has been to drive these folks into the Republican party. Thus, being a Republican is now associated not only with modern racism but with OFR as well. Since party identification tends to persist throughout the life cycle, this realignment may affect our politics long after Obama's presidency. Tesler also predicts an increase in overtly racist rhetoric, since it is now useful in mobilizing the Republican base.

Deadbeats for Self-Reliance

Binyamin Appelbaum and Robert Gebeloff had an article in Sunday's New York Times suggesting that some of the people who are most in favor of severe cuts in government domestic spending are also highly dependent on the federal safety net. This chart—actually a scatterplot—by Aaron Carroll supplements that article. It shows that the residents of conservative states get more help from the government than liberal state residents. The chart requires some explaining.

The vertical axis measures government help. “Transfer receipts” refer to all the payments made by the federal government to individuals, such as social security, Medicare and unemployment benefits. It is divided by total personal income to yield a percentage, so .20 on the chart means that the people of that state derive 20% of their income from government programs. As you might expect, those states that are higher on the vertical axis are those states whose per capita income is lower.

The horizontal dimension refers to how liberal or conservative the state was in a 2010 Gallup poll. “Conservative advantage” refers to the percentage of self-identified conservatives minus the percentage of self-identified liberals. From left to right, the “conservative advantage” increases. The size of the circles represents the population of the states.

The trend is obvious. The more conservative the state, the more government “handouts” its citizens receive. Including all 50 states yields the same result. Of the 25 most liberal states, 16 of them pay more in taxes than they get back. But 23 of the 25 most conservative states are on the receiving end of wealth redistribution. This also translates into voting behavior. Seven of the ten most dependent states voted for McCain in 2008, while only two of the ten least dependent states did.

There are two tempting ways to interpret this chart. One assumes that red state citizens are aware of the inequity from which they benefit and are hypocrites. The Times article reports that when the conflict between their attitudes and their behavior is pointed out to some of them, they rationalized that they “deserve” their government benefits while most other recipients do not.

Another possible interpretation is the false consciousness argument from Thomas Frank's What's the Matter with Kansas? It suggests that poor state residents are confused and, without realizing it, are voting against their economic self-interest. In effect, they have been duped by the Elephant Party, which appeals to them with ads about abortion and gun control, but once in office, uses their electoral advantage to enact economic policies which primarily benefit the wealthy.

(Of course, you could also argue that anyone who voted for the Jackass Party in the last 20 years has been duped, since once in office, the Jackasses have rather consistently done the opposite of what they promised to do during the campaign.  See "The Mourning After.")

A new analysis of wealth and voting patterns by Andrew Gelman adds some interesting data to the mix. He finds that, in all 50 states, the higher your income, the more likely you are to vote for the Elephants, as would be predicted by self-interest. However, the higher the per capita income of a state, the higher the share of its vote that goes to the Jackasses. That is, there is something about the culture of the poorer states, perhaps relating to social rather than economic issues, that persuades their people to vote for the Elephants.

Furthermore, the relationship between income and party preference is stronger in the poorer states than in the richer ones. As a result, a high percentage of low income people in all states—rich and poor—vote Jackass. But in the poorer states, the middle class and especially the rich, vote strongly Elephant, while in the richer states, there is a higher percentage of Jackass voters among middle and higher income people.

In other words, most of the poor people in all the states are not confused. They vote in their self-interest. But in the higher income states, people who are not themselves poor vote in a way that appears to benefit the poor. (If anyone votes contrary to their self-interest, it's rich people from Connecticut.) In the low income states, upper middle class and rich people vote in a way that is more consistent with their economic self-interest and that is harmful to the interests of the poor. Here's a presentation by Gelman:

Of course, all of these data are correlational, and correlation does not mean causation. A wealthy Mississippian may not necessarily have voted for McCain because he wanted poor people to starve. He may have been motivated more by an opposition to gay marriage, and the effect of his vote on the plight of the poor may have been incidental. Voting is complicated.