Trends in the Religious Unaffiliated, the “Nones,” by Age

Last week I posted a figure showing the rise of the religiously unaffiliated over the past 40 years. Today I am breaking it down by age, for this helps us to understand better what is happening. As shown last week, rates of the unaffiliated rose rapidly in the 1990s and more slowly since. Two general models for this rise would be be 1) it’s happening primarily among the youth, and they remain unaffiliated as they age or 2) it’s happened to all ages over time, though youth were, and continue to be, less affiliated with religion.

In this first figure, I plot the percentage of Americans who report being unaffiliated with religion by three age groups: the young, ages 18-29, the middle-aged, 30-49, and us old folks, age 50+. Using data from the General Social Survey, I was able to go back to 1972, and to smooth out the lines, I collapsed data into five-year intervals (e.g., 1970-74, 75-79, 80-84, etc…)

As you can see, rates of the religiously unaffiliated have increased rather steadily in all age groups. Yes, the young are more likely to be unaffiliated the the middle-aged or the old, but it’s been this way all along. Something that I did’t appreciate until I broke down the data this way is that the religious revival in the 1980s seems to have taken place primarily among the young. The strong rise in the unaffiliated in the 1990s happened mostly among the young, as they lost the religious gains of the 1990s.

In terms of relative trends by age, the above figure is a little difficult to interpret because each group started at a different baseline. So, to standardize starting points, I set each group’s 1972 level to equal “1,” and then I plotted relative changes from that shared reference point. As such, a score of “2″ would indicate that the rate of being unaffiliated had doubled.

Once again, the percentage of being unaffiliated increased in each group, but relatively speaking, it’s increased most among the middle-aged and the elderly. In both the percentage of the unaffiliated more than tripled, compared to the 2.5x increase in the young. There is some lagged effect, as the elderly are catching up the middle-aged in the past decade, but overall, the rise of the religious nones is something that spans all age groups. Thus it’s a societal-wide change more than just an age or generational change.


If Physicians Acted Like Sociologists (or, Why We’re Mostly Irrelevant)
Denominational Change in American Protestantism since 1972
Are We Losing the Young? Evangelicals by Age Since 1972 (Part II)
How Christians Vary in their Awareness of God
  • Marlin J Harris

    This is very interesting, but honestly not too surprising. The reason I say this is because especially now with society, religion is being taken out, it’s not allowed in our schools, Christians are being killed or put in jail for their faith, ect. The media and the government are putting God on the back-burner, so society is falling into that way of thinking/living too.

  • Larry Chouinard

    It would be interesting to chart the divergent reasons for church departure among the various age groups. Where are they similar and different in their reasons for departure. Thanks for sharing your survey.

    • Bradley Wright

      Not sure how to do that, but it would be very interesting!

    • abb3w

      There was a GSS module in 1988 that asked people about changes in religious affiliations, but most people never change, so the total N was only around 350 or so. That’s including changes from one Christian denomination to another, and still too small to get statistically significant results for breakdown by cohort and reason.

      You’d probably have to do a massive oversample to get data allowing serious analysis. Say, use the entire KnowledgeNetworks panel to find the ~20% unaffiliated, cull out the ~5% who were raised that way, and ask the remaining 15% more extended questions. That’s likely in the $100k cost neighborhood, though.

  • oph

    Isn’t this neglecting the cohort effect? E.g, the 20 year olds of the 1980s are the 30 year olds of the 1990s. If each new cohort is less religious than the one above it, then even if there is no change in people as they get older, the age trend will still resemble the lines we see above. Not saying all the change is about cohorts, but my messing around with the data suggests that’s most of it (my apologies if I misunderstood your claim here, or if you’ve already covered cohort change in a previous blog post).

    Also, maybe of interest, is that the GSS 2012 cross-sectional data was just released today, in case you were in the mood to update your graphs with the most recent wave.

    • abb3w

      This is in fact the case. I usually use recodes on COHORT and RELITEN for looking at this. There is only slight shift in time within cohorts (EG: the very early 1970s tended slightly less religious; the 2000s may also), but considerably between cohorts (EG: circa 30% of millennials are now unaffiliated, but only circa 10% of boomers). It’s not all about cohorts, but that looks to be the 1000 pound gorilla.

      More quantitatively — the number of irreligious in a cohort appears to approximately fit a logistic curve, with time constant circa 27 years and midpoint expected circa the 2007 Cohort.

      Unfortunately, Berkeley’s SDA interface doesn’t have the GSS-2012 up yet. However, those deeply determined to look at the latest data can download that SPSS data file from NORC, and play with it in R.

  • Craig


    I propose a theory that these “nones” are not leaving but are being pushed out of their respective churches by homophobes. Whether the pastor formally announces that gay people are not wanted in his church or the social conversation subtly informs everyone that gays are not welcome, the people leaving organized church are being told to choose church or gay rights and they are choosing gay rights.

    How do we go about testing this theory? Is there a way to track and compare the change in opinion about gay marriage rights in America vs the rates of people leaving church? Is that a valid way to see if there is a causal relationship?