Skip to content

Judith Rich Harris, correlation, and causation

by Michael Nielsen on February 18, 2004

In 2001 Judith Rich Harris published “The Nurture Assumption”, which, from the many favourable mentions I’ve seen in a variety of contexts, seems to have had a considerable influence on our thinking about how human beings develop.

I haven’t yet been able to get a hold of Harris’ book, but I did just find an essay by her in a collection of essays put together by John Brockman, “The Next 50 Years”.

The argument Harris makes in the essay can be summarized as follows:

(a) Genetic factors play a large (though far from exclusive) role in controlling human development.

(b) Most studies in developmental psychology assume that it is the environment that determines human behaviour.

(c) Those studies therefore fail to control for genetic effects.

(d) Those studies are therefore essentially useless.

It’s a fascinating argument: simple, yet if correct, devastating for an entire academic field.

(To avoid any confusion in attribution, I should say that I don’t know who originated this argument, although it is clear that Harris is but one of many contributors.)

Everyone knows, at some level, that correlation does not imply causation, but it is difficult to avoid the fallacies that can result when one is not mindful of this.

It occurred to me when talking with my partner, Jen, this evening that I don’t have a nice vivid example of this point ready at hand. So we made one up that I think illustrates the point memorably.

Suppose one observes top basketball players, and notices that they all tend to wear very expensive basketball shoes. Is one justified in assuming that more expensive basketball shoes cause someone to be a better basketball player?

In some small measure, the answer is yes. But it doesn’t matter what you put on my feet, I’m never going to jump like Michael Jordan.

Indeed, marketing in general seems to be based largely on the fact that, at some level, in many cases subconscious, we easily confuse correlation with causation.

From → General

  1. Well said, Michael. I especially like your last point which is something that I had subconsciously realised but never so nicely codified like you have here: “marketing in general seems to be based largely on the fact that we easily confuse correlation with causation.”

  2. Mary Messall permalink

    Again, this seems like just the right context for my geeky reaction:

    Nuture? Pah. Nature? Pooh.

    Our genes are a fractal equation, and our childhoods provide the initial conditions. Where we will be at any given point in life is chaotic, unpredictable, a feedback process wherin what we do depends entirely upon what we’ve done.

    I think I actually want a better math metaphor in which even the *form* of the equation depends upon the point you’re at.

    Anyway, I choose to believe that who you are depends mainly on the choices that you make. I also have geeky physics arguments about the existence of free will.

  3. A Crawford permalink

    full disclosure: I’m prejudiced regarding developmental psychology.

    That said… in their defense. a) Genetic factors (GF) are a necessary element in all cases. But in a given case where the sum of result is attributable (and/or ascertainable), X=GF+EF+a-z, the contribution of GF to X is a range between a minimum (moot, noise, bias) and a defined maximum.

    b) Some studies consider both Environmental Factors (EF) and Genetic Factors (GF)… (c) Cases that do not consider (GF) do not necessarily have a flawed outcome because some (GF) could be at or below the minimum. Yet:

    d) As all cases where (GF) are not considered have at least a minimum relation to the result, X, those lacking (GF) could be suspected of committing a fallacy of Exclusion.. i.e. Important evidence which would undermine an inductive argument is excluded from consideration.

    Asking that all relevant information be included is called the “principle of total evidence” (for example, Specs for all equipment used). It is not sufficient simply to show that not all of the evidence (raw data) was included; it must be shown that the missing evidence has a likely or necessary potential of changing the conclusion (weaken or void an inductive argument).

    Pretty tough question for a Psychologist: “Is there a reasonable potential that Genetic factors are contributing in any significant way to your results? Are you qualified to determine genetic factors? You are, after all, not a biologist and don’t have a background in genetics. Knowing this did you extend the scope of your initial queries?


Comments are closed.