The sleep of the innocent

I sleep fairly well — as I’ve gotten older, I sleep less, but most nights I go to bed, read quietly for an hour or so, and within minutes of turning out the lights, I’m out. Then I’ll typically wake up around 5:30am. I get between 6 and 7 hours of solid sleep most nights, which is about what I need. That’s my pattern, I’m sticking with it.

Clearly, the reason I sleep well is because I’m an atheist — you know, unconflicted by storms of Catholic guilt or Protestant sanctimony. At least, that’s the implication of a study. After all, when it’s called a “study”, you know it’s serious business.

A new study out of Baylor University finds that 73% of atheists and agnostics sleep at least seven hours a night, compared to only 55% of Baptists and 63% of Catholics. Atheists and agnostics also reported fewer difficulties falling asleep at night. The findings held even after controlling for details like age and socioeconomic status.

Gosh, it fits my personal experience as an atheist, so I get that little buzz of confirmation. I don’t have any personal experience sleeping as a Baptist or Catholic, but I still get a little buzz at the idea that the Others are tossing and turning at night. And now, when my religious friends (if I had any!) show up at work tired and red-eyed, I get the smug satisfaction of being able to tell them they ought to become an atheist so they can get a good night’s rest for a change.

If I were an idiot.

I took a look at the study. They’re just mining a bigger data set for correlations, which is not a kind of science I care much for. In other words, there was a survey that asked a large collection of semi-random questions of religious and non-religious people, and then afterwards they fished for anything that might show a difference. It could maybe be useful if they followed through and figured out what caused the difference. But they don’t; this is just a cold wet plop of an observation from a data set.

Another thing that always bugs me is that when a study is reported in the popular press they always strip out important qualifiers that at least the original work includes.

Religious affiliation was associated with sleep duration, but not in the predicted direction. Atheists/Agnostics (73%) were significantly more likely to report meeting consensus sleep duration guidelines than religiously-affiliated individuals (65%), p< .05. For example, Atheists/Agnostics reported better sleep duration than Catholics (63%, p< .01) and Baptists (55%, p<.001). Atheists/Agnostics also reported less difficulty falling asleep at night than Catholics (p=.02) and Baptists (p< .001).

Notice the word I highlighted: these are the results of a self-reported survey. They don’t necessarily mean what you think they mean. It could mean that atheists are very insecure and like to lie positively about their health and confidence. Go ahead, Baptists! You can read it that way! Even if I know in my heart that you are tormented by your god-belief.

That’s the only thing that jumped out at me about this trivial and mundane study. It’s a great example of how the press likes to file the serial numbers off a paper: they report “atheists sleep better” when the actual paper says “atheists report that they sleep better”. In this case, they’re making a change that’s favorable to the godless, but it’s also what they do with studies that go the other way. This paper even begins with a statement of the common presupposition!

The psychology of religion literature indicates that religious engagement is beneficial to physical and mental health.

I don’t trust that interpretation any more than I do the suggestion that atheism is beneficial to your sleep.

Also, it’s a Templeton funded study.


  1. nomadiq says

    Alternate interpretation. Atheists are generally wealthier and economic security would make one sleep better at night.

    Methinks the authors would ‘test’ this idea by more self-reporting on economic anxiety instead of running some kind of controlled experiment.

  2. says

    Well more important than the self-report angle is that this is an unadjusted bivariate association. There are many other differences between atheists and religious people, including for example level of formal education, affluence, and probably age. If they didn’t adjust for any of that this is completely meaningless.

  3. dorght says

    I tried to correct my lead programmer once that mining a large dataset for correlations is not proof of anything. It didn’t go well. Since this was in the world of criminal justice joined with mental health care all kinds of spurious correlations could arise. I liked liked that our program really helped divert people from jail to mental health diversion court, but I really hated when management was trying to sell database mining results (thankfully they never did).

  4. consciousness razor says

    cervantes, #2:
    They do say this after the part PZ quoted above:

    The effects persisted when controlling for age and were particularly evident in members of African American congregations. Perceptions of getting into Heaven were significantly higher in participants who obtained better sleep duration, p<.05, but interestingly, such beliefs/perceptions were unrelated to difficulty falling asleep at night, suggesting that better sleep may lead to these perceptions rather than vice versa.

    The Baylor study it’s based on did ask people to self-report on tons of other potentially relevant things, regarding their health, emotions, stress, whether a loved one died recently, amount of exercise, drug use, what they estimate their social class to be, satisfaction with their financial situation, closeness to family/friends/neighbors, romantic partners, sexual activity, and on and on.

    But strangely, they don’t bring up any of that stuff in the summary of the results above. Not that most of it would be very useful, given that we’re talking about self-reports. Still, if this the game they’re playing, then did they just not consider any of it or what’s that about?

  5. birgerjohansson says

    OT -Speaking of correlations, for half a year I have hoped to see a study of how vitamin D deficiency correlates to poor outcomes of Covid-19. Now there is an interesting study coming out of Cordoba, Spain. It is only based on 74 patients, yet the correlation of poor outcomes is striking. I would recommend PZ and other persons in the danger zone use supplemental vitamin D, it is dirt cheap after all.
    BTW, if you want to sleep as badly as the catholics, here is something to think about at night.

  6. DonDueed says

    Don’t you see, PZ? This study proves that goD is real! The faithful get that extra energy boost from their invisible friend, so they can function fine with less sleep.

    Meanwhile we lazy, tired atheists have to sleep extra hours to make up for our supernatural deficit. QED!

  7. PaulBC says

    The succubi don’t bother showing up to disturb your sleep when they know Satan has your soul already. They’d never make quota if they did. Atheists and agnostics are also far less likely to succumb to disease spread by all manner of demon effluvia.

    Hope this helps!

  8. billseymour says

    dorght @4

    I tried to correct my lead programmer once that mining a large dataset for correlations is not proof of anything.

    When working on a computer system to track the use of Postal money orders for money laundering, a new fellow in our group suggested all the ways that we could scan money order data looking for fraud of various kinds; but the Postal Inspector who was our “customer” on the project put the kabosh to that immediately. We absolutely do not spy on our customers willy-nilly. The clerk who sold the money order had to file a report, and an Inspector at headquarters had to look at an actual image of the money order. Only after two H. sapiens agreed that the money order looked suspicious would our system click in.

    “If this were my employer’s opinion, I wouldn’t be allowed to post it.” — Norman Diamond

  9. PaulBC says

    I would like to get 8 hours, but I often get 7 or less (but about 7). The main thing is that I like to be out of bed for an hour or two before anyone else in my household is up, like right now. My wife and kids aren’t early risers so it’s not a competition. I rarely have any trouble getting to sleep and when I do there’s a specific cause, such as work or personal stress. I can sleep through loud noises (like the very unusual Bay Area lightning strikes last month). Since doing 100% work from home, I’ve been more inclined to take naps in the afternoon as inclined.

  10. robro says

    Also, it’s a Templeton funded study.

    And Baylor is a Baptist school, indirectly affiliated with the Southern Baptists. It’s mildly surprising that they would release a report that someone might construe as negative about Christians. On the other hand, perhaps they see a positive: Christian don’t need to sleep as much because they rest in the arms of the lord, or some such blather.

  11. PaulBC says

    robro@11 Or they don’t indulge themselves with slothful rest. Work ethic and all that. Amount of sleep strikes me as a neutral finding. Some may actually be proud of getting as little sleep as possible (and not just religious people but workaholics).

  12. says

    It’s not the “sleep of the innocent.” It’s the “sleep of those with a fully integrated conscience.” It’s easy to see how one might easily lose an hour of sleep a night with extra worry over whether one’s morals for the day were meeting someone else’s standards — especially “someone” who isn’t human and hasn’t spoken directly for (at least) a couple of millenia, and therefore it’s pure conjecture on that person’s interpretation of modern society and satisfactory ancestry.

    Just remember, guys: All of the leading figures in That Book were Semites.

  13. says

    This kind of fishing expedition shouldn’t be called a “study”. Large data sets are great for inspiration, to be sure, but you ought to actually test your hypothesis before publishing.

  14. says

    I’ve always been an atheist and my bad conscience mostly involves undone household chores, half prepared lessons and an eternal fear to have missed an appointment. Our crazy crafting ideas. Over the years I’ve developed special techniques to clear my mind of those things so I can fall asleep.

  15. says

    Also, earth’s 24 hours day does not correspond with my biorhythm. Ideally I’d get 9 hours of sleep, six hours awake, one hour nap, 10 hours awake…

  16. PaulBC says

    bcwebb@18 That sounds like an argument for bringing back human sacrifice!

    (Note: no it doesn’t, but it is the first thought that leapt to mind.)

  17. says

    Not a big surprise that a dataset with 11k people had sleep self reported. I’m guessing there are other measures on the questionnaire that could be used to test whether atheists are just lying to make themselves look healthier etc…
    Of course there are caveats to the study and who knows about reproducibility, but science happens in stages and the effects religious beliefs have on sleep is kind of interesting 🤷‍♂️….at least interesting enough that they tested for it and pz commented on it.

  18. unclefrogy says

    It would be nice if my sleep patterns were in any way regular I would be grateful but unfortunately they are not. Sometimes I get up to 9 to 10 hours other times it is 4 to 5 hours. these days i deal with stress with varying degrees of success. Not sure how different it was when I was younger and was not as aware of what was going on and had much less to think about.
    here I am any way
    uncle frogy

  19. birgerjohansson says

    After 40 years, the Alien xenomorph still makes appearences in my dreams. So I do not sleep well.

  20. says

    Like many in my family, I’m a chronic napper. I really can only sleep for about four hours at a time but I do it twice a day. It frees up a weird time just past midnight that I have to do something with. Also makes long term relationships somewhat problematic.

  21. Kevin Karplus says

    Did they say what multiple-hypothesis correction they made for all the hypotheses they checked? This looks like p-hacking to me.

  22. slithey tove (twas brillig (stevem)) says

    makes sense to me (sorta) Catholics are indoctrinated to feel guilty about everything. To always keep wondering “what they could have done” to get past every error they see, regardless of their involvement.
    I say this having been raised in an RCC household, and spent most of my adolescence extracting myself from the cult.

  23. thrymskvida says

    I’ve been an atheist since birth and also a terrible sleeper since birth — interesting to see I buck the trend proposed by totally legit sources!