It might simply be their age. Maybe they will be more likely to identify with a religion when they're older. Alternatively, there could be an uptick in the numbers of people who are leaving religion - for good.
It's actually pretty difficult to untangle the statistics to work out what's going on here. Philip Schwadel, a sociologist at the University of Nebraska, has used a couple of newly developed statistical techniques to try to do just that, using data from the General Social Survey (which has been surveying Americans since the early 1970s).
What he found was that all three effects seem to play a role. He found that, across all generations since around 1990, there has been a sharp increase in the numbers of people reporting that they have no religious affiliation.
But, surprisingly, younger generations aren't more likely to drop out of religion than they were before. In fact, people born to a religious family in the 60s and 70s are no more likely to switch out of religion than were people born before 1945.
Not so for people born in the period 1945-1960. They are more likely than older generations to switch out of religion. Clearly, growing up in the Hippy generation had its effects!
About one quarter of the increase in non-affiliated young adults can be explained simply by the fact that more and more American kids are being raised in non-affiliated families. This is the snowball effect. When the Hippy generation grew up, they passed on their lack of affiliation to their kids - who were joined by other people who are continuing dropped out of religion at the normal, background rate.
The big question now is what will happen to these young non-affiliated. Based on earlier generations, you might expect a fair number of them to rejoin a religious identity as they age. But will this happen to the Millennial generation? Time will tell!
Schwadel, P. (2010). Period and Cohort Effects on Religious Nonaffiliation and Religious Disaffiliation: A Research Note Journal for the Scientific Study of Religion, 49 (2), 311-319 DOI: 10.1111/j.1468-5906.2010.01511.x
This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.
(1) You speak of Schwadel's data since 1990 but the graph only shows to 1984 and is labeled to 2006. Can we see the rest
ReplyDelete(2) "used a coup newly developed statistical techniques"
Can you tell us simply what "NEW" techniques can mine data differently than the old ones?
Sabio, (1) the graph shows birth cohorts. So the people in the final column on the graph, born 1980-1984, reached adulthood just in time to be surveyed in the 2003 GSS. The graph showing the rise in non-affiliated is output from the model. I was going to put that up but decided it was too complicated to explain, which brings me too...
ReplyDelete(2) not really, because I don't fully understand them myself! Basically the first uses a variation of hierarchical regression, in which birth cohort and time period are controlled for before assessing the effects of other factors.
The second uses principle component analysis to strip out the common variation that's caused by the links between age, period and birth cohort so that you can analyse each of these separately.