Skip to content

Belief in hell is associated with lower crime rates

I remember attending a talk a few years ago by my political science colleague John Huber in which he discussed cross-national comparisons of religious attitudes. One thing I remember is that the U.S. is highly religious, another thing I remembered is that lots more Americans believe in heaven than believe in hell.

Some of this went into Red State Blue State—not the heaven/hell thing, but the graph of religiosity vs. GDP:

and the corresponding graph of religious attendance vs. GDP for U.S. states:

Also we learned that, at the individual level, the correlation of religious attendance with income is zero (according to survey reports, rich Americans are neither more nor less likely than poor Americans to go to church regularly):

while the correlation of prayer with income is strongly negative (poor Americans are much more likely than rich Americans to regularly pray):

Anyway, with all this, I was primed to be interested in a recent study by psychologists Azim Shariff and Mijke Rhemtulla, “Divergent Effects of Beliefs in Heaven and Hell on National Crime Rates.” It’s published in PLoS One, which seems to be the new Arxiv: they’ll publish anything, and studies there seem to be able to get publicity.

Shariff and Rhemtulla find the interesting result that, in a cross-national comparison of countries with data from the World Values Survey, the prevalence of belief in heaven correlates with higher crime rates, and belief in hell correlates with lower crime rates. These patterns remain after controlling for some country-level variables including indicators for the dominant religion in the country. This result surprised me at first—if asked, I would’ve had the vague idea that belief in hell is associated with violent, unsettled lives—but it makes sense from simple psychology: belief in a pleasant afterlife reduces people’s inhibitions while belief in a punishing God could deter one from crime. One can also imagine these effects at an aggregate level whether or not they appear for individuals. A society with widespread belief in hell could have a more punitive culture in which crimes are more strongly frowned upon, etc.

Unfortunately the open-data movement has still not reached universality. All the authors’ data sources are freely available but the numbers have not been released with the published article; thus if I actually want to check the numbers or do my own analysis I’d have to put in the work of finding the data, downloading them, and reformatting. (I’m just as bad in my most of my own published research: I analyze publicly-available survey data but I don’t always have the processed data in a convenient form for others.)

I thought about this after studying Shariff and Rhemtulla’s (only) graph:

You can click to see the plot full-sized. I was stuck (a) trying to figure out which countries are which, and (b) puzzling over the crime numbers. I started with the US, which appears to have above-average crime numbers (check) and only 10% more belief in heaven than hell. I’d thought the difference was more than 10%, but I’m probably misremembering John Huber’s talk. Other countries: it appears that Venezuela and Guatemala have higher crime rates than South Africa. That surprised me, but hey, there it is on Wikipedia. But what about “SE” and “NO”—are these Sweden and Norway? There’s no way they have higher crime rates than the U.S. So I’m not sure what’s going on here.


  1. Bill Mill says:

    I’d guess that they’re using ISO 3166 country codes, under which SE and NO are just as you guessed:

  2. keith says:

    Wouldn’t belief in hell be more for more traditionally religious people, usually older and more conservative, less fighty?

  3. Anonymous says:

    I wonder what controlling for religion means?

    I could look at paper, but off the bat control for Catholicism is not the same as controlling for Christianity, or for religions that include Hell.

    Is it asymmetric belief in H&H given religion has notion of H&H that matters, or judeo-christian religion say versus other.

    Also worry ecological inference

  4. Mikael says:

    Sweden at least has pretty high crime rates, partly because of very high reporting. I don’t know if that applies to Norway as well.

    • Andrew says:

      When it comes to homicide, Sweden and Norway are way below the U.S. But maybe they have high victimization rates for important crimes such as snowmobile-jacking and fishing without a license. Given them a little fear of hell and those Nordics would be all law-abiding, just like Americans.

  5. Thomas says:

    As I am about to move to Sweden, I wanted to check this out. Apparently, violent crime rates are very low in Sweden (, in fact among the lowest in the world. Ditto for Norway. Other crimes may indeed be reported differently than in other countries. For example, Sweden is said to have one of the highest rape rates in the world, about double the reported rate for Zimbabwe. This appears to be an artifact of a difference in what is considered rape, as the Assange case shows. The authors of the cited papers should do their analysis for different categories of crime. Murder rates may be more comparable than say pick-pocketing across the world.

    • Andrew says:

      Data + Stata = Paper!

    • Nameless says:

      You are correct that there are different kinds of crime, and one needs to know what they meant by “crime index”, and that reporting is a big issue. Murders are most likely to be reported in every country, but they constitute a very small fraction of all crime. In the United States, various forms of theft account for more than 80% of all reported crime. Within the U.S., correlations between violent crime rates and property crime rates are often weak. Among major cities,

      the correlation between murder rates and vehicle theft rates is R^2=0.439, between murder and burglary, 0.183, and between murder and larceny, 0.005.

      Murder rates aren’t perfect, either. I recall reading that Australia, New Zealand, and the UK have more overall violent crime per capita than the U.S., but the U.S. has a 3-5 times higher homicide rate. The difference is because the U.S. is a handgun-happy place (more handguns than adults) but there are very few handguns in AU/NZ/UK. Therefore, violent crime in the U.S. is much more likely to have a fatal outcome.

  6. Sander says:

    The Crime Rates for Sweden and Norway cannot be right. They are way higher than those recorded for Russia and the US. There is something really dodgy going on there. Also one should bear in mind two things, Asian countries have very low crime rates (for a whole set off cultural reasons) and almost no one believes in either hell nor heaven so they all cluster in the low 0 region of the graph. Latin America has really high crime rates for a bunch of other reasons and in general really high differences between heaven and hell beliefs (because people there are pretty christian, which really seems like the main driver of that difference). Include region dummies for those two, and put in correct crime statistics for Scandinavia and Eastern Europe and I doubt the story will stand.

    • jrkrideau says:

      Re “dodgy”

      It may be much higher reporting of crimes or a difference in definition or who knows what. Some years ago I was looking at homicide and other crime figures for the USA and Canada and, somewhat to my surprise, Canada had a considerably higher reported rate of assualts.
      Going from memory took a reading of some Statistics Canada report footnotes to realise that while Canada reported three levels of assault, the lowest level was not in the US data.

      I began to get the idea that international comparisons of crime rates was a dicey problem at best. Note that it cannot be done but it is always going to be a bit messy and needs caveats on some of the data.

      The general consensus seemed to be that nationl level homicide rates, at least in most OECD countries were pretty solid though even there weird things can happen.

      For example, Australia’s murder rate jumped drastically one year — not that there were that many more murders that year but the bodies of a long-time serial killer’s victems were found and all counted in that year.

  7. MIchael says:

    Something odd in your own first graph, as well. When and by what measure was Japanese GDP per capita not less than USA?

    • Andrew says:


      We’ve discussed this on the blog, not the specific case of Japan but in general there is no unique way to compare countries’ GDP. We found values for Russia that varied by a factor of 5. I think what we did to make the graph was to choose a single source, recognizing that it was not perfect. Similar problems arise when trying to adjust state incomes for cost of living within the U.S. (another topic that I’ve blogged on a couple times).

      • MIchael says:

        I’m sorry I missed that discussion. It’s true there is no uniquely “right” price index (which is what you would need to have a unique way of comparing GDP) but there is pretty much a consensus among international economists (I am a minor one) that for most purposes we want to look at one of a few rather similar options. If there is no reason not to, go with Purchasing Power Parity based numbers, as in the International Comparison Project at U Pennsylvania. To get a factor of five difference, or Japan richer than the USA, I think you must have been looking at something that converted nominal GDP to dollars using market exchange rates. These should converge with PPP rates in the long run (we proclaim) but in any given year they can be way off due to speculator beliefs about future monetary policy and such, which have no real bearing on how much the average person can buy with his income.

  8. Jeremy Miles says:

    PLOS One won’t publish *anything*. But if the methods are sound, and the results justify the conclusions, they will.
    (I’m an associate editor at PLOS One, I’ve handled 9 papers and rejected 2).
    In addition, PLOS One gets a statistical review of every article (or at least, every article I’ve been involved with) and the statistical editor gets paid ($100, I think).

  9. Nameless says:

    Looking through raw crime data in Sweden vs. the United States. In the study, Sweden gets a high crime score because it has significantly higher numbers in categories “assault” and all property crimes (“theft”, “motor vehicle theft”, “burglary”, “robbery”).

    The “assault” difference looks like a reporting artifact. U.S. numbers are for “aggravated assaults” only, not including “assaults that do not involve the use of a firearm, knife or cutting instrument, or other dangerous weapon and in which the victim did not sustain serious or aggravated injuries”. Swedish data are almost certainly for all assaults.

    Differences in property crimes, however, still look real, even after digging to the bottom:

    Pickpocketing: 0.08/1000 in the US vs. 4.78/1000 in Sweden
    Shoplifting: 3.43/1000 in the US vs. 6.63/1000 in Sweden
    Motor vehicle thefts: 2.39/1000 in the US vs 3.65/1000 in Sweden
    Bicycle thefts: 0.67/1000 in the US vs 6.95/1000 in Sweden

    • idiot says:

      Does this mean the crime rate in Sweden is much more scarier than the crime rate in USA? I would think that it is much easier for someone to steal your property than to murder you, so if property crimes increase, that means you are more likely to be a victim of said crime (assuming you actually have property, so that excludes the very poor) than if you just walk down the street and get shot.

    • StefanP says:

      In Sweden you must report your stolen property to the police to get compensation from your insurance and the very report should be submitted to the insurance company. I guess it should be the same in other countries as well, but there may also be between-country differences in the likelihood of that actual police report being entered into the database. Most of these cases cannot be solved with reasonable effort so there could be an incitement for these smaller crimes not showing up in the crime statistics. Or maybe I have seen too much of The Wire as well as being too quick to put on the Freakonomics(TM) hat!

      The police force in Sweden is under quite some fire for having a very low frequency of resolving (smaller) property thefts. But there may also be reporting differences. The crime with an almost 100% rate of being solved is police murder. There is some interesting bias! One may wonder how that solution rate is in other countries?

      • Nameless says:

        It works the same way in the States. Thefts have to be reported, the insurance company pays up, and everyone forgets about it.

        I don’t know what the likelihood of a report being entered into statistics is.

        It is a fact (a curious one) that pickpocketing is basically dead in the United States, for multiple reasons. In Europe, it is kept alive by immigrants from places like Romania.

        With regard to vehicle thefts, I suspect that many cars are stolen from Sweden and end up in Eastern Europe. There’s also a strong bias towards stealing two-wheeled vehicles. Almost a third of “motor vehicle thefts” in Sweden is motorcycles and mopeds. There is a large number of bicycle thefts. In the States, these do not occur frequently simply because very few people ride mopeds and bicycles. It could be a cultural thing and/or a consequence of higher urbanization in Sweden.

    • Mikael says:

      Interesting. I have lived in Sweden for about 35 years now and I can’t recall hearing a single story about pickpocketing from my friends – but “everyone” has been a victim to it in Paris or Barcelona. Maybe it’s really common in certain parts/cities.

      Bicycle theft – now that’s another matter. Anyone who has lived in Uppsala will have had their bike stolen multiple times (usually by drunk students).

      • Nameless says:

        According to the statistics, about a quarter of all pickpocketing incidents occur in Stockholm’s inner city.

  10. Snarkyxanf says:

    The axis there is “heaven believers minus hell believers”, which means someone who believes in both and someone who believes in neither both count the same.

    Apparently universalists are the cause of crime.

  11. Fr. says:

    I’m pretty sure this graph also says that Catholicism (green+orange) drives the crime rate up. In other words, I smell something spurious.

    • Nameless says:

      No, it only says that Christianity is correlated with the heaven-minus-hell metric and with crime rates. To decorrelate them, authors of the study did a statistical regression on crime, belief in heaven, belief in hell, and other factors. Only beliefs in heaven and hell still stood significant, Christianity was factored out.

      But I was surprised that the top three countries on the heaven-minus-hell metric were all Roman Catholic. I thought that hell was always a much more important feature in Catholicism than in Protestantism.

  12. Eli Rabett says:

    Think a about where AL or MS vs NH or VT would fall on that chart.

  13. Ben B says:

    I wonder if we should read anything into the precise concentration of swing states (PA, OH, MI, FL, VA, &etc) in the graph of religious attendance vs. GDP for U.S. states. I wonder if there is anything interesting to be found in the change in these measures over time and voting tendencies.