Gene Expression

Over at Why Evolution Is True Greg Mayer wonders:

I also recalled that the percentage of religiously unaffiliated had gone up noticeably from 1990 to 2008, and that another survey found the percentage was higher among young people. What could have happened so that younger people, growing up in the 90s and 00s, would be less religious? And then it occurred to me: 9/11. Something finally happened which gave religion a bad name. This was forcefully expressed at the time (here, here, and here) by Richard Dawkins.


The the fact that the % of Americans who aver “No Religion” has increased greatly in the past 20 years is a surprise to many people, because naturally it isn’t as if Richard Dawkins is organizing revivals. Like the decline in crime in the 1990s this is a trend which social scientists have no good explanation for. The crash in church affiliation in the 1960s was accompanied by concomitant cultural changes. Though the 1990s were certainly vibrant, with the explosion of the internet, I don’t think people will speak of a “90s generation” in a way that we talk about the 60s generation. From what I have heard the cultural gap between 1960 and 1970 was far greater than the cultural gap between 1970 and 1990.

In any case if Greg Mayer is correct we should see a shift in religious affiliation after a specific date. The American Religious Identification Survey does not support the hypothesis that 9/11 was a hinge of history. Here are the proportions of those with “No Religion” by the survey years:

1990 – 8.2%
2001 – 14.2%
2008 – 15.0%

These data suggest that the “action” occurred during the 1990s. The General Social Survey allows us to drilldown to a more granular year by year scale:

(I put the 95% intervals, and limited the data sets to “whites” so as to eliminate the confound of changes in ethnic proportions)


Here’s the table with detailed results….

YearNoneLow boundHigh Bound


  1. #1 agnostic
    August 14, 2009

    Changes in any index of identity are never “due to” single events, although these can play exacerbating roles. When something rises and falls, and then rises and falls again, we’re dealing with a process that can generate oscillations of some kind.

    When a pendulum starts to swing the other way, it is pure error to ask what event happened at that moment, or just before, which caused it to move back. It’s the laws of motion that cause this, not a particular event, like an angel blowing the pendulum really hard in the other direction.

  2. #2 Tony P
    August 14, 2009

    Read your Freakonomics. The crime drop in the 1990′s is posited to have been due to the legalization of abortion. Fewer kids growing up in poverty and the like.

    Very interesting way to look at the issue.

  3. #3 razib
    August 14, 2009

    The crime drop in the 1990′s is posited to have been due to the legalization of abortion.

    *posited* it isn’t accepted as the clear explanation.

  4. #4 Gray Gaffer
    August 14, 2009

    “The crime drop in the 1990′s is posited to have been due to the legalization of abortion”

    Uh, by what mechanism and why the delay? Roe vs Wade happened in 1973. I do not recall it resulting in any drop in the poverty levels. Those seem to have just always been on the increase since, oh, the 50′s? or forever? Anyone got a graph for that?

    I would note that the apparent drop in the 60′s was not actually a swing towards atheism. I was there. It was a swing toward spiritualism. To alternatives to religion, to Hindu, to Zen Buddhism, to Tibetan Buddhism, to Sufism, to Gaia cults, to Wicca, etc etc etc. To Transcendental Meditation, to various personality cults of the Maharajas and other gurus (not all Eastern). “No Religion” was the answer that covered that in the polls, which did not have boxes for the reality.

    Much woo was born then too. All fueled by wildly enthusiastic interpretations of the effects of various psychoactive substances. And sex. And more sex. And Rock ‘n Roll*. Even people who were already not part of the Established Religions like Christianity or Judaism woke up and took part. So in fact religiosity increased, not decreased, just not in the direction of accepted affiliations.

    20 years later, in the 80′s, the backlash started, as we woke up again to the irrelevance of it all. And the dangers. And friends started dying. Then there was a swing to atheism.

    * “Sex, Drugs, and Rock’n'Roll” was not just an amusing slogan of the times. The three together were the driving force of the alternative generation. All the spirituality and other such stuff descended from the release from the oppressive Protestant mores and the permissions that that implied. The drugs knocked one sideways, the rock posed questions, one went into search mode. And the power of the three was personified for me in the psyche of one of my gfs of the time: any two induced a strong desire for the third for her. Which could be embarrassing at a rock concert. With strangers. Operant conditioning, I believe the term is.

    cut to: Raul Julie (Adams Family dad) saying with an evil grin “but I’m much better now”. Which movie was that from?

  5. #5 Larry
    August 14, 2009

    Though the 1990s were certainly vibrant, with the explosion of the internet, I don’t think people will speak of a “90s generation” in a way that we talk about the 60s generation.

    I think the Internet revolution of the 90s is just as significant as the social/sexual revolution of the 60s. Maybe you’re right and people won’t talk about the 90s generation — after all access to information does not have the sizzle of access to sex and rebellion against authority, and there’s no Bob Dylan singing about the Internet. Nonetheless, the fact that people now have orders-of-magnitude better access to information revolutionary. My own rapidly progressing deconversion would not have happened without the information I found through Google and Amazon.

  6. #6 agnostic
    August 14, 2009

    The abortion explanation is wrong. Steve Sailer’s covered it in enough depth, with refs to the academic lit, for homicide rates. And the same is true for the concomitant decline in all forms of child abuse (see Finkelhor’s free reviews at his university website).

    There was a decline in crime starting in the mid-1930s, after a wave of increasing crime starting around 1900. Was that due to abortion? Obviously not. We have a rise, fall, rise, and fall. That requires a single model that produces oscillations, not four separate and unrelated ad hoc causes for each rise and each fall. That would be stupid.

  7. #7 Joshua Zelinsky
    August 14, 2009

    It seemed to me that by and large the immediate response to 9/11 was more religious fervor rather than less. I a junior in highschool at the time and there were all sorts of religious services organized by students both at my school and in the general community. However, some of that seemed to be people taking comfort in ritual more than an actual increase in belief.

    The conventional wisdom is that when catastrophes occur people look to religion for answers. It may very well be that the long-term impact of 9/11 will be to reduce religious fervor but with first a short-term bump. (That is complete speculation).

    If I had to make a guess at any event that caused the downward trend in the data in the 90s it would be the fall of the USSR. There are two possible reasons for this: First, there’s less reaction against being areligious since there’s no longer the big-bad-atheist evil empire. Second, with less of a worry about the possibility of humanity being wiped out any second in a global nuclear exchange there’s less need for the comfort of a guiding hand or the comfort of life after death. It might be interesting to see what those numbers look like in 1995 or so. If this hypothesis is correct we should see a sharp increase in the “No Religin” percentage (or percentage answering similarly) early in the 1990s.

  8. #8 Brian Schmidt
    August 15, 2009

    Anything Steve Sailer says is pretty suspect, IMHO.

    The “more abortion, less crime” hypothesis could be tested pretty well using Romania as an example. Abortion went from illegal to widely available in 1990. I found some preliminary data in 2006 for Bucharest that was somewhat supportive of the hypothesis:

    Much more definitive data should be available now.

  9. #9 TGGP
    August 15, 2009

    Agnostic, it is alleged that the figures for crime in the first decade or so of the 20th century severely undercounted homicides, so there was less of an increase to the peak than we thought.

  10. #10 Douglas Knight
    August 15, 2009

    Greg Mayer’s hypothesis seems psychologically backwards to me. If you substituted “nationalism” for “religion” no one believe it for a second.

  11. #11 mihai
    August 16, 2009

    The “more abortion, less crime” hypothesis could be tested pretty well using Romania as an example. Abortion went from illegal to widely available in 1990

    The drop in crime in Romania starting from about 2002 has another explanation. In 2002 or 2003 (don’t remember exactly) visa requirements for the UE contries were lifted and a massive emigration (2 mil people) began, mostly too Italy and Spain. A not so small part of those people were the local crime scene. The effects are clearly visible in Bucharest, for example. The streets are very safe at night and you can safely use the buses or trams (in the 90′ they were full of pickpockets). Also, most of the abortions were made by middle class maried women and by students. The legalisation didn’t significantly alter the behavior of the lower class women.

  12. #12 Brian Schmidt
    August 16, 2009

    Thanks mihai – that’s something to keep in mind, but it wouldn’t necessarily ruin the “experiment.” As long as the ability to emigrate isn’t affected by whether you were born before 1990, then it shouldn’t matter.

    Actually, it’s probably somewhat harder for juveniles to emigrate than adults, so that might somewhat skew the results artificially against the “more abortion, less crime” hypothesis.

  13. #13 toto
    August 17, 2009

    Sailer’s counter-argument is notable for including its own potential counter-counter-argument.

    Essentially the crime wave peaked in the 90s, pretty much when the first kids of the Roe vs Wade generation came of age. Sailer points out that this seems incompatible with the abortion explanation.

    But then, he himself observes that this peak corresponds to the “crack epidemic” of the 90s, and the carnage that ensued. He attributes much of the later drop in crime to the fallout of this event – essentially positing that it scared the shit out of the survivors, and knocked some law-abidance into them.

    But this immediately raises the possibility that this “epidemic” was actually a purely external event, occurring against a background of falling crime (which may or may not be a result of abortion). The crime data alone cannot suffice to distinguish between the two possibilities, because they would produce the same graph. He doesn’t seem to acknowledge this possibility.

    That’s also why modelling social trends with purely self-contained differential equations is a waste of time. External, unpredictable events do occur. How could you predict the modern crime statistics of Europe, without including a (necessarily ad hoc) component for the two world wars?

  14. #14 J.J.E.
    August 18, 2009

    and there’s no Bob Dylan singing about the Internet

    Well, he’s no Dylan, but there is Jonathan Coulton signing now!