Gene Expression

The rise of the irreligious Left

Barry Kosmin at CUNY has published the results of three surveys of American religion since 1990. These “American Religious Identification Surveys” (ARIS) were done in 1990, another in 2001, and finally in 2008. One of the major findings of the ARIS has been the rise of those who avow “No Religion”. Looking through the data it is also clear that aggregating nationally understates some of the local changes. In 1990 47% Vermonters were non-Catholic Christians (i.e., Protestants). In 2008 29% were. In 1990 13% of Vermonters had No Religion. In 2008 34% of Vermonters had No Religion! In fact, No Religion has a plural majority in Vermont, with 26% of the population being Catholic. This is a much bigger shift than nationally. In Kosmin’s book One Nation Under God, which drew upon the 1990 survey results, he noted that though the Northeast has a reputation for being relatively secular, it is in fact highly confessionalized in comparison to other regions, such as the Pacific Northwest. This isn’t true anymore; much of New England has experienced a wave of rapid secularization and disaffiliation. If current rates of secularization continue Vermont may become the first minority non-Christian state. It was only 55% Christian in 2008.

I am going to explore these data with scatterplots and maps. All the plots have 1990 data on the X-axis and 2008 data on the Y-axis. I also look at the 1988 vs. 2008 presidential race, and attempt to see if there is any connection between secularization and political change.

Here’s a scatterplot showing % with No Religion in 1990 & 2008 by state:


R-squared 0.6 means that 60% of the variation in 2008 can be explained by variation in 1990. This is a reasonably large value, and indicates a real relationship, but it also shows that there’s bee some resorting. No state has seen a decline in those with No Religion, but the states to the top right of the plot have seen the largest shifts. In particular, above the trendline are the states of northern New England, which have secularized a great deal. Below the trendline you have western coastal states, such as Oregon, which were relatively secular in 1990 and so did not change as much as Vermont. As you can see, you have a cluster of states, disproportionately Southern, which show less change.

The next scatterplot compares absolute change (e.g., a 5 point change in No Religion) and relative change (a 100% change in No Religion):


Here you can see that Vermont, New Hamphsire and Delaware show both large absolute and relative changes in those with No Religion. By contrast, North Dakota shows a large relative change, but a modest absolute change. That’s because North Dakota started at a very high level of confessionalization, so any shift would show as a large relative change. Oregon, California and Arizona are on the bottom left of the chart because these were already rather secular states, so the relative and absolute changes are not particularly great (they’re below the trendline in the first scatterplot).

Here are some maps which show the geographic clustering….



Now I want to see how (non)religious change tracked political change. I am going to look at the % vote for Republican presidential candidates in 1988 and 2008, and compare them to ARIS 1990 and 2008. First, how do the 1988 and 2008 presidential elections compare?


I was a little surprised at the low R-squared here; but realignment has progressed it seems. Now let’s compare the percentage with No Religion and voting for the Republican in respective years.



This is not too surprising, over the past generation the Republican and Democratic parties have polarized when it comes to social issues. Not only are there more individuals with No Religion today than in 1988, but there’s also more variance.

There’s a lot of unaccounted for variation when it comes to voting for the Republican candidate in 2008 vs. 1988. Vermont, for example, became a very Democratic state, whereas in 1988 it voted for George H. W. Bush. By contrast, West Virginia has shifted from a solid Democratic state to a Republican leaning state, at least on the national level. I calculated the residual off the trendline from the 2008 vs. 1988 scatterpline. Vermont is nearly 14 points below the trendline (-14) and West Virginia 15 points above the trendline (15). Oregon is around the trendline. I compared this to the change in the proportion of those with No Religion in absolute terms.

First, a map which shows the trend of the shifts from ’88 to ’08 as measured by the magnitude of residuals off the trendline:


And now the scatterplot:


Some of the variation can be accounted for by change in the proportion of those with No Religion. In particular, northern New England and Delaware have becoming far more Democratic as they’ve become more secular. Arkansas is an example of a state which has had little religious change, and has become more Republican. Finally, you have a state like California, which hasn’t become that much more secular, but has become more Democratic. I think this is a function of the ethnic change within the state, not religious change. The new Latino immigrants aren’t particularly secular, but they’re very Democratic, in relation to the non-Hispanic whites.


  1. #1 Tony P
    January 24, 2010

    Actually I thought RI was about 20% no religion. The more telling factor in RI is when you look at Catholics. They went from 62% in 1990 to 49% in the last survey and dropping since.

    New England overall runs at about 20% which is quite remarkable, this being the cradle of the United States and all.

  2. #2 razib
    January 24, 2010

    it’s 19 percent in my data. i guess i should post that in a table too….

  3. #3 darwin
    January 25, 2010

    Razib, does the survey questioning sort out people who answer with a religion out of cultural loyalty or identification, but who are athiests or non-believers? I think many Jews and Catholics would answer on this basis (other religions as well, although I personally know fewer examples there). If the questions do not sort out such people, degree of unbelief is very much understated.

  4. #4 Spike Gomes
    January 26, 2010

    No Hawaii? By the latest rough estimates, we haven’t had a Christian majority in awhile. They don’t really poll here. We’re too demographically weird. Frex, 5-10% of the population calls themselves Buddhist.

  5. #5 razib
    January 26, 2010


    the N = 201, but you see a safely majority xtian state. more so than vermont.

  6. #6 Leonard Bernstein
    January 26, 2010

    A recent survey of incoming students at Tufts University, Medford, MA, asked, amongst other items, the religious affiliation of the student. An amazing 30% of the students claimed no religious affiliation.

    Further, the Student Secular Alliance ( has been experiencing unprecedented growth on college campuses nationwide.

    Politicians are beginning to take note of these trends which are likely to enter more and more into the political equation.

    Even members of Congress are beginning to announce their secular, non-religious status, Rep. Pete Stark, D-CA, as an example.

    It may soon come to pass that we will have a country with a true “wall of separation between Church and State”.

  7. #7 JakeS
    February 9, 2010

    The first plot seems pretty convincing, but the rest look like shotgun plots to me.

    The hypothesis is convincing, given that the Republican Party has, over the studied period, officially taken leave of its senses and is now being run in large part by ayatollah-wannabes. That naturally drives away non-believers more than those who share some or all of their theology. But this data can’t really be said to provide independent support for the idea.

    - Jake