On August 24, 2018, President Trump warned 100 evangelical leaders who had been summoned to the White House that the fall’s midterm elections are “very much a referendum on not only me, it’s a referendum on your religion.” White evangelical Protestants had overwhelmingly–77% of them– voted for candidate Trump in 2016 (versus 57% of white mainline Protestants), while the religiously unaffiliated had voted overwhelmingly against him–only 24% for Trump. A year and a half later, white evangelicals and the unaffiliated were virtually unchanged in their rates of heavily supporting versus heavily opposing the president.
Religion and politics have been intertwined through much of American history, especially since Catholics starting moving here in large numbers. Americans’ party allegiances tended to follow their religious allegiances. Seemingly new to this century is the extent to which the reverse is true, that politics is driving religious identity.
Since this possibility was first discussed by some of us about 15 years ago (1, 2, 3), this reversal of mover and moved has become increasingly apparent. A few weeks after Trump’s declaration to the assembled evangelicals, the election-predicting website 538 posted a story titled “Americans are Shifting the Rest of their Identity [sic] to Match their Politics,”drawing on an as-yet unpublished study I discuss below.
Two recent developments in this story lead me to update earlier posts (4, 5, 6): In the academic world, research on how politics shapes religious expression has boomed. And in the real world, the mutual influence between Caesar’s realm and God’s realm has tightened.
Old Battles
Americans’ politics has long been connected to their religious identities. The 19th century and the early 20th century were full of faith-based controversies and religiously-based voting. Early on, sectarian battle lines formed over, for example: laws enforcing a Christian Sabbath (see earlier posts on struggles over Sunday mail and Sunday baseball); the Bible’s place in public schools (and whose version of the Bible; see this post); rights of religious dissidents such as Mormons, Jehovah’s Witnesses, and especially Catholics; abolition; and again and again, alcohol. The disputes divided Catholics from Protestants, Protestant denominations against one another and sometimes within one another, and more secular Americans against more religious ones. And often these conflicts spilled into street violence. (A seasonally appropriate example is the Yom Kippur riot of 1898.)
Court rulings in the 20th century built up the wall between church and state in ways that reduced the power of the state–and thus the importance of politics–in realms of faith. If, for example, public schools could no longer (since 1962) stipulate a daily prayer, elections could no longer determine whose prayer is stipulated. Still, the wall is not that high and has many gaps, and the connection between religion and politics has tightened since the 1970s.
New Battles
One way to see this tightening is in the graph below which I drew from the General Social Survey (GSS). It shows white Americans’ political party preferences from 1972 to 2016 according to their religious affiliations. For each group for each year, I calculated the difference between the percentage who declared themselves Republicans or leaning Republican minus the percentage who declared themselves Democrats or leaning Democratic. A dot at +.20 means that the group leaned Republican that year by a difference of 20 percentage points; a dot at -.60 means that the group leaned Democratic that year by 60 points. Five groups are shown: fundamentalist Protestants (light blue), other Protestants (orange), Catholics (gray), “Nones” (no religious preference, dark blue), and Jews (red). The lines summarize the trends displayed by the dots.[1]
Here’s what I see:
* Overall, white Americans moved from leaning Democratic in the 1970s (years of fallout from Watergate and the Vietnam War) to leaning Republican in the 1980s and early 1990s (the “It’s Morning in America” years of Ronald Reagan).
* Then, the trends diverged. White Protestants, especially fundamentalist Protestants, became increasingly Republican, while Catholics, Jews, and especially “Nones” became slightly or substantially more Democratic.
* In the 1970s, white fundamentalist Protestants were about 10 percentage points more Republican-leaning than white Nones were. In the 2010s, they were over 60 points more Republican-leaning.[2] The 2016 points, by the way, precede the election of Trump.
What happened? One thing that happened was that the culture wars associated with the rise of the Religious Right in the 1990s re-entangled faith and politics. Women’s issues, pot, gay rights, sexual freedom, lifestyle choices, and abortion, of course, mobilized religious groups and anti-religious groups.
But what also happened, it increasingly appears, is that people (and/or their children) shifted their religious identities to better fit their positions on the politically divisive issues. They moved from one of the lines in the graph to another, especially to the Nones.
Nones grew from about 7% of GSS whites in the 1970s to about 22% in the 2010s, while non-fundamentalist Protestants shrank from 38% to 26% and Catholics stayed constant at about 28% (constant most likely because Hispanic immigration compensated for defections; Jews were constant at about 2.5%).[3] Interestingly, the fundamentalist Protestant share of the white population grew from about 23% in the mid-1970s to about 30% in the mid-1980s and fell back to about 23% in the mid-2010s.
Some of these patterns are the result of varying birth rates and the rise of immigration. But much seems to be about people changing religious allegiances. And the argument is that those changes in religious identities were driven in part by political concerns. Let’s turn to a few of the new academic studies that support this argument.
Cause and Effect
* NYU political scientist Patrick Egan, in a just-released paper, replicated and extended analyses Mike Hout and I conducted in 2014 on GSS panel data. The GSS re-interviewed most of their 2006 respondents again in 2008 and in 2010, asking the identical questions each time. Egan examines whether respondents’ political stances in 2006 (liberal vs. conservative, Democrat vs. Republican) predict any changes in religious identification between 2006 and 2010. Indeed, they did. Respondents who called themselves liberal Democrats in 2006 were about 4 percentage points more likely to say they were of “no religion” in 2010 than they had in 2006, while conservative Republicans were about 6 points less likely to claim no religion. Conversely, liberal Democrats became 4 points less likely and conservative Republicans 4 points more likely to call themselves either Protestants or born-again Christians. (Strikingly, political identification also affected changes in the probability of respondents labeling themselves as Latino or as LGBT.) That a mere four years could lead to such measurable changes makes compelling a conclusion that decades of political polarization have substantially altered Americans’ religious identities.
* David Campbell of Notre Dame and his colleagues reported two studies. One, like Egan’s, makes use of a panel survey, here with four waves. The authors estimate the extent to which changes in religious identity at one time are associated with changes in political party the next time or vice-versa. Their results also show that changes in party identification more strongly influenced the odds of reporting oneself a “None” than vice-versa. The second study was an experiment. Subjects whose religious identities had been identified a week earlier read (fictional) presentations by political candidates. Some candidate profiles invoked religion and others did not. Then, the subjects were again asked about their religious identities. A notable proportion of subjects who had been assigned readings about highly religious Republicans now declared themselves “Nones.” “The close association of religion and the Republican Party,” wrote the researchers, “creates cognitive dissonance among Democrats. Many Democrats resolve the dissonance by becoming Nones” (or, more accurately, by declaring themselves Nones).
* Michele Margolis similarly combines survey data analysis and experimental work in her 2017 paper. (She describes her work in a New York Times essay.) Margolis adds to the discussion the argument that politically-driven change in religious identity is specific to a stage in life, the stage when young adults start raising children and contemplate whether to re-engage in the religious tradition in which they grew up. In her experiment, people who had two weeks earlier described their levels of religiosity read flyers that either did or did not discuss politics and were then re-asked questions about religion. Those who had been exposed to political tracts were likelier to change answers when then asked about religion: Democrats reported themselves as less and Republicans as more religious than they had before–but only if the subjects had young children.
One survey Margolis uses followed a cohort of high school students from 1965 to 1997. Both young Democrats and young Republicans attended church less often after they left their parents’ homes, but the Republicans were likelier than the Democrats to return to the fold, so that the attendance gap between Republicans and Democrats widened. Results from another panel survey, conducted between 2000 and 2004, show that the party difference in church attendance widened in that period–but only for respondents with children at home. Margolis also makes the historical point that as the generations that have experienced the intensification of the party-religion connection since the 1970s have matured into parenthood, their growing numbers have made that association all the stronger in the overall population.
* Finally, political scientist Paul Djupe of Denison College and colleagues looked closely at who left specific congregations since the mid-2000s. Using three different panel surveys, they found that the increasing political connection to religion led evangelical Christians who felt politically out of tune with their fellow congregants–recall, their fellow Christians were moving right–to leave their particular churches. That exit, in turn, may have set the stage for some of those to finally deny identification with their previous faith tradition entirely.
These studies–and there are probably others–reinforce what would have seemed a bizarre notion to observers of the severe religious-political conflicts of the 19th century, that political disputes can alter religious identities.
Update, Dec. 19, 2019:
Is there a change brewing?: Headline in Christianity Today, a major evangelical publication: “Trump Should Be Removed from Office.”
~~~~~~~~~~~~~~~~~~~~~~~~
Notes
[1] I look only at whites to set aside for present purposes the special case of blacks, who tend to be religious but since the 1930s almost totally Democratic. I exclude the few who answered the religion question as “other” and those who answered the political question as independents not leaning either way. Fundamentalists are Protestants whose denominations are coded that way by the GSS. I used a quadratic function to summarize a trend if it provided a substantial improvement in R-squared over a linear function.
[2] These numbers are based on comparing the 1970s averages to the 2010s averages of the Y-axis values.
[3] Calculations based on 5-sample moving averages.
MADE IN AMERICA 