How different are men and women? Are they from two different planets?
In the cleverly-titled The Distance Between Mars and Venus, the authors argue that personality-wise, the differences between men and women have been underestimated by previous studies because they used simplistic statistics.
Traditional studies of gender and personality have given some men and some women a personality quiz, and calculated the average male and female scores on the different aspects of personality.
When you do this you find that there are differences, but that the standardized effect sizes are fairly small, which means that there is a lot of overlap. Even on measures where men score above women on average, lots of men score below the female average, and vice versa, like this:
Traditional studies of overall gender differences have looked to see the differences between the average man and woman on each personality aspect, and then averaged the differences on each scale to get an "overall difference" score. Which comes out as fairly small.
The authors of the new paper say that this approach fails to capture the true difference and they give a helpful analogy of why:
Consider two fictional towns, Lowtown and Hightown. The distance between the two towns can be measured on three (orthogonal) dimensions: longitude, latitude, and altitude. Hightown is 3,000 feet higher than Lowtown, and they are located 3 miles apart in the north-south direction and 3 miles apart east-west.
What is the overall distance between Hightown and Lowtown? The average of the three measures is 2.2 miles, but it is easy to see that this is the wrong answer. The actual distance is the Euclidean distance, i.e. 4.3 miles – almost twice the "average" value. The main novel argument of this paper is that if you calculate the distance (technically the Mahalanobis distance) in 'personality space' between men and women then you get a larger value than if you just average the differences on each measure.
The paper also uses a couple of other methods that increase the effect sizes, namely using 15 different personality measures instead of the more common Big 5, and adjusting the differences upwards to take account of the fact that quizzes only imperfectly measure underlying 'latent' personality traits.
I don't want to get into the debate over how valid the underlying data are (a 1993 sample of over 10,000 American adults, used to standardize the 16PF questionnaire). There are lots of technical comments here. I'm going to focus on the distance method.
It's a very interesting approach and certainly raises questions about merits of the old approach, which when you think about it, does seem a bit crude. But I'm not sure that the average person is talking about distance in a hypothetical space when they talk about "personality differences".
As an analogy, consider the dog breeds Labrador and Golden Retriever. These are regarded as being pretty similar kinds of dog. On any given feature, the average differences are small, at least compared to the diversity of other breeds. They're roughly the same size, much the same build, coat type etc.
They are distinct breeds. This surely means that when you take all of the differences together, they define distinct regions of "dog space" (which has dozens or hundreds of dimensions), with little or no overlap.
Yet they are still regarded as similar."Similar" and "distinct" are not mutually exclusive. In fact, isn't the definition of 'distinct yet similar' that two things separate in some kind of feature-space, but don't differ much on any one measure?
So I would say that these data show that, while men and women may be distinguishable in personality, they could still be similar. This is something of a semantic point but not "merely" semantic: it changes the interpretation of the numbers.
J. S. Hyde, who is most associated with the view that gender differences are small, makes a similar (or do I mean distinct?) point in her comment on the paper:
The gender difference found is along a dimension in multivariate space that is a linear combination of the original variables transformed into latent variables...[but] the resulting dimension here is uninterpretible. What does it mean to say that there are large gender differences on this undefined dimension in 15-dimensional space created from latent variables? The authors call it global personality, but what does that mean?
Her questioning of what the direction along which men and women differ means, is (I think) the same question I'm asking about whether it disproves the idea of "similarity", in the ordinary sense of the term.
Finally, take a step back and the whole debate seems a bit circular because, by definition, "personality" means "things that differ between individual people". Things we are have in common aren't even in the picture. Two groups could differ in personality space but still be very close in the much larger space of "possible creatures". There's no personality trait for 'being human'.
Del Giudice, M., Booth, T., and Irwing, P. (2012). The Distance Between Mars and Venus: Measuring Global Sex Differences in Personality PLoS ONE, 7 (1) DOI: 10.1371/journal.pone.0029265