Stephanopoulos used two sources for his analysis: first, the massive database of public opinion surveys covering 2,074 questions over a 25-year period (1981-2006) compiled by Martin Gilens for his book “Affluence and Influence;” second, he combined state level exit polls from between 2000 and 2010 with an index of state policy liberalism, thus allowing him to compare the preferences of different groups with the outcomes of policy. He controlled for group size by multiplying a group’s support for a policy by its share of the population. Like Gilens, Stephanopoulos only compared policies in which preferences between the two groups being studied differed by more than 10 points.
As shown in the charts below, Gilens’ dataset demonstrates convincingly that the preferences of women, people of color and those falling into low- and middle-income brackets have almost no effect on policy. A flat line indicates that policy preferences have no influence on policy, a downward-sloping line means that a group’s policy preferences have a negative impact on policy. An upward-sloping line indicates that a group’s preferences have a positive impact on policy — that is, they get what they want. This dataset is incredibly useful, because rather than examining broad ideological measures, it directly asks: Do the people get what they want?
The answer is: Yes, if they’re rich white men.
The negative line for women is particularly troublesome. For more than half of the United States population, supporting a policy actually means policymakers are less likely to pass it. As Stephanopoulos explains,
“As male support increases from 0 percent to 100 percent, the odds of policy enactment rise from about 0 percent to about 90 percent. But as female support varies over the same range, the likelihood of adoption falls from roughly 80 percent to roughly 10 percent. When men and women disagree, then, stronger female backing for a policy seems entirely futile.”
This is absolutely insane.
While the Gilens dataset only measures policy outputs at the federal level, the second dataset examines how opinions and policy correlate at the state level. The charts below document how the ideology of different groups interact with the ideology of state policy.
As you can see, whites and Latinos have positive and statistically significant influence on policy, while blacks do not. This means that if blacks in Mississippi are more liberal than blacks in Alabama, that won’t affect the relative policies of each state. And, once again, women have no influence on policy at all, nor do low-income people. Indeed, Stephanopoulos writes that “state policy liberalism actually decreases from about 1 (or roughly Ohio’s policy set) to about -1 (or roughly New Hampshire’s) as the ideology of those making less than $30,000 varies over the same range.” That is, poor people’s preferences are negatively correlated with policy in their state.
Much of what Stephanopoulos finds meshes with other research on the subject: Other studies (particularly the work of John Griffin and Brian Newman) find that people of color have divergent preferences and that those preferences aren’t well represented. However, Stephanopoulos finds much larger gender gaps than other research has suggested. Griffin and Newman, for instance, found that gender gaps in representation were rather small. However, their method was based on how frequently women had policy “wins,” meaning that the policy they supported was enacted. This means that their methodology can’t recognize “democracy by coincidence,” which Stephanopoulos’s work suggests may be occurring.
In another work, Griffin and Newman with Christina Wolbrecht find that much of the gender gap could be explained by the fact that Republicans tend to oppose policies women support. Indeed, much of the work on class and race on representation findsthat parties can serve as partial explanations for representation gaps by mediating preferences. The Republican Party has become “an efficient patronage machine for whites and the top 1 percent of U.S. earners.” Because the Republican party acts for the benefit of exclusively white rich men, whereas the Democratic party must juggle a broad and diverse coalition, the party system can give rise to representation gaps. However, parties are not the sole explanatory variable, because there is strongevidence that the Democratic party has shifted to the right in order to accommodate the increasing influence of the wealthy (and the declining influence of unions). In addition, Griffin and Newman note (see chapter 5) that while Democrats represent blacks and Latinos better than Republicans, they still favor the preferences of whites.
