January 25, 2017

What Diverse Countries Really Look Like

We are often told that ethnic diversity is a strength. Given this, we should expect that diverse countries are, in various ways, better than homogenous ones. Let’s take a look at how accurate this expectation is.

To look at this question empirically, I used data on life expectancy, life satisfaction, and national wealth from the 2016 World Happiness Report and data on ethnic diversity from Alesina et al. (2003). Conceptually, ethnic diversity is defined as the probability that two people randomly selected from a nation will not be of the same ethnic group.

As it turns out, the more ethnically diverse a nation is the less happy its population tends to be. Across 129 nations, the correlation between diversity and national life satisfaction is -.40 which is a moderately strong negative correlation by conventional standards.

Life Satisfaction.png

Ethnically diverse nations also tend to be poorer. The correlation between diversity and national wealth across 129 is -.55 which is a strong negative correlation by conventional standards.

National Wealth.png

Finally, we have health. The more diverse a nation is the lower its life expectancy tends to be. Across 130 nations, the correlation between diversity and life expectancy is -.66 which is a very strong negative correlation. (There was life expectancy data for one more country than there was for life satisfaction or wealth).

Life Expectancy.png

Diverse nations tend to be less healthy, less happy, and less wealthy, than homogenous ones. Of course, these are just correlations, but correlations are evidence (though not proof) of a causal relation. Here is a brief account of one plausible hypothesis about why these variables are related which has supporting evidence from various studies:

Experimental, longitudinal, and correlational, evidence shows that ethnic diversity damages how well people get along with one another. A person’s social life, in turn, is thought to be a key determinant of how happy they are. This may be why ethnic diversity negatively correlates with national life satisfaction.

Experimental and correlational evidence also shows that group productivity is damaged by ethnic diversity. Correlational evidence further shows that distrust in others, which ethnic diversity creates, increases the degree to which people favor government control over the economy. These are both plausible mechanisms by which diversity might hamper national wealth.

This decrease in national wealth could, in turn, also cause a further decrease in national life satisfaction.

We don’t need to posit anything more to explain life expectancy. National poverty and unhappiness are not good for a person’s health, and this may explain why diversity negatively correlates with life expectancy.

Advocates of diversity must argue that these correlations are in some way misleading and present a more compelling explanation for them than I have here in order to make a convincing case for allowing our nation (or your nation) to become more diverse.

Of course, they won’t. Instead, they will try to silence rational debate with cries of racism. But facts don’t care if liberals call them racist, and the fact of the matter is that diversity looks a lot more like a national weakness than a national strength.

Facebook Comments
  • Emil Kirkegaard

    Add IQ as a control and you will see large reductions in the validity of these.

    • That is probably true, but IQ co varies so much with some of these variables, and due to causation running both ways as well as numerous confounding variables, that I would be skeptical of usefulness of using IQ as a control.

      • Emil Kirkegaard

        But that’s the point. You cannot really say much about the effect of diversity if you don’t control for obvious confounds.

        By the way, diversity has a non-linear relationship to outcomes in the US when analyzing counties. https://openpsych.net/paper/12

        • pymotes

          It’s important to note, as you have, that IQ has a relationship with the independent & dependent variables here — but it’s also important to note that if we control for IQ here we dismantle reality: it would be like saying “if we pretend these groups are all the same on IQ”… which takes us a step away from reality. altho it would be nice to see the squared semi partial correlations & overlapping venn diagrams, but for most folks, that would be tl;dr:)

        • Frontierland

          The US White population has decreased precipitously from about a 90% White homogeneous nation in 1970. Given that you’re modeling the outcomes of a very short time span, and given the demographic shift has only just begun to overshadow the host White population, I’d like to see your diversity outcome models in 10 or 20 years.
          I’m sure there will be little doubt as the effect.

          • jz95

            And I have little doubt that you, like the majority of commenters/writers on this site, have 0 white children.

        • Controlling for confounding variables can be misleading. It is true that national IQ causes variation in health, wealth, and possibly life satisfaction, and that a bi-variate analysis may attribute this variation to diversity. However, controlling for national IQ will also remove all the co-variation of IQ and these national outcomes. This includes not only variation explained by IQ causing variation in national outcomes but also variation explained by differences in national outcomes causing differences in national IQ and co-variation between IQ and national outcomes which is caused by confounding variables. Because of this, controlling for IQ could be misleading if we are trying to measure the causal impact of diversity on national outcomes.

        • It is also worth noting that the impact of diversity on economic and psychological outcomes is supported by experimental evidence.

        • Also, there are several lines of research to suggest that ethnic diversity itself may have a causal impact on cognitive ability.

    • Segregate

      Imagine if we use this argument to control for environment+every gene except ones coding for melanin and aesthetics unique to each race, when looking at crime done by race. We can say of course that behavioral traits such as impulsiveness and IQ are confounds as justification, but only if we say that we wish to prove that it is the visibly identifiable features unique to each race that cause crime in the first place, which no one claims that they do. It is the difference in behavioral traits like IQ that cause difference in crime in the first place.

      Then, if we wish to measure the negative effects of ethnic diversity, it is only realistic to ensure a representative population of each race in the analysis. If we control in the effort for it not to be representative, I don’t see how that’s useful. Who’s to say that “ethnic diversity” is independent of every behavioral trait? It’s really the very point.

      • Emil Kirkegaard

        Many people do claim that diversity has negative effects that not due to the compositional effects. I did not find that in my analysis, except in so far as one will claim that the ~100% White areas were evidence of the bad effects homogeneity and not say just due to being rural or whatever. I did not examine the issue in detail because it’s a minor issue to me.

    • NuclearBlackMetalKampf

      For example, a country is much worse half Somali than half Croat.

      More high-crime, low-trust groups = more misery. Reality be Rayciss and Colonialist and Sheeit.