Predicting self-categorization 1: Comparative fit

Theoretically this is a post that you all have been waiting for with bated breath. If self-categorization is as fundamental as we say, and is as implicated in as many important social phenomena as we say, then knowing which self-categories will become psychologically active, as well as when, is going to be of enormous significance.

To give an example, we have already spent a fair bit of time on leadership, where we have explained that a person’s leadership credentials are determined by the extent to which they are central to other peoples’ psychological ingroups, or ‘social identities’. We have also made the case that if you are interested in leadership development then your first step should be to “spend time getting to know the social identities and social identity environment of those [you] wish to lead” (and that to fail to do so is to risk getting it wrong). What then could be more vital than understanding the things that predict social identification?

First though, let’s establish why we are talking about this at all. I mean, is it really that complicated? Isn’t it just similarity? That is, we put things in the same category if they are similar and things in different categories if they are dissimilar. Or in self-categorization terms, shouldn’t we just expect people just to view similar people as ingroup members, and dissimilar people as outgroup members?

‘Similarity’ isn’t at all a crazy thing to say. In fact, it’s probably what I would have said before the social identity approach got its hooks into me. It’s wrong though. Or rather, it doesn’t work as a solution to the particular problem at hand. I’ll explain…

Let’s start here: In the below figure, are the two points close together or far apart?

I promise that I am not just being obnoxious. There is a point to this.

I promise that I am not just being obnoxious. There is a point to this.

Well, what do you think? Hard to tell? It is for me too. In fact, I would argue that the question can’t be answered. There is some critical information missing here.

Speaking of can’t be answered, take a look at any two things around you and ask yourself if they are similar or not. Right now in front of me I have a book and a watch. Are these two things similar? They share some features, but certainly not others. Do they belong in the same category? There are also some sunglasses and earphones on my desk. Are they similar? They both go on my head. But they are worlds apart in a myriad of other ways. Do these belong in the same category or not?

The point is that similarity doesn’t cut it as the predictor of category membership. We need another answer. Of course, we have one at the ready; an answer that is spelled out in the social identity approach. Or more specifically, in self-categorization theory,[1][2] which is one of the two constituent theories of the social identity approach:

Self-categorization is predicted by an interaction of three factors. These are ‘comparative fit’, ‘normative fit’, and ‘perceiver readiness’.

Don’t worry. We are not going to try to cover all three of these today. We are instead going to start with comparative fit, in what is akin to a ground up approach.

Comparative fit

Let’s go back to our two points; the ones that are neither close together nor far apart. This time, however, there will be another point in the scene.

See, I was only partly being obnoxious.

See, I was only partly being obnoxious.

Now this is much easier. ‘A’ and ‘B’ are clearly far apart from one another. Meanwhile, ‘B’ and ‘C’ are clearly close together. This is a depiction of comparative fit, also known as ‘structural fit’, operating in the simplest scenario possible.

The idea of comparative fit is that stimuli will be classed together if the average distances between those stimuli are observed to be smaller than the average distances between those stimuli and other stimuli within the frame of reference.[3] This can also be communicated in what is known as the meta-contrast ratio, which turns the above into something that can be expressed mathematically. It does this by making inter-category distances a numerator and intra-category distances a denominator:

‘nd’ is the number of relevant dimensions; ‘nx’ is the number of members of some category X; ‘ny’ is the number of members outside the category X; and the ‘x’ and ‘y’ values are the positions of a member of one of or the other category on a particular dimension. We owe Craig McGarty for the exact formulation here.[4]

Not a maths person? Don’t worry about it. The underlying idea is actually straight forward. What we are talking about is the way that our minds’ use both comparisons within stimuli and comparisons between those stimuli and other stimuli to decide what sort of categorization scheme is most meaningful. In our ‘A’, ‘B’ and ‘C’, example our mind has used literal distances to determine which points go together and which are apart. Elsewhere though these “distances” could be anything. Any feature or combination of features can create distance between things in a conceptual sense. Take these shapes for example…

Something other than my cutlery for variety sake.

Something other than my cutlery for variety sake.

First, let’s look at the left hand side. On the one hand, each of these shapes is different from the others, or in our language there are conceptual distances between all the shapes operating on a number of dimensions. Some are larger than others and some are greener than others. On one dimension, some of the shapes have more ‘squareness’ (or alternatively ‘triangle-ness’) than others. This is not to say that half the shapes are perfect squares and half the shapes are perfect triangles;  simply that there are shapes that are closer to squares than they are triangles and other shapes that are closer to triangles than they are squares.

This is comparative fit, and the various conceptual distances created by ‘degrees of squareness’ means that we are able to create a categorization scheme where overall the distances within the categories are smaller than the distances between the categories. Those categories, of course, are ‘square-like’ and ‘triangle-like’.

Now let’s look at the right hand side. Most of the shapes are the same, except now that categorization scheme has some competition. This is because the frame of reference has changed. Or in other words, we are now faced with a different array of stimuli where more things are present. New conceptual distances among the stimuli are introduced on new dimensions. An alternative categorization scheme emerges; we can now see ‘curved shapes’ alongside ‘straight shapes’.

Do you see it? Moreover, are you experiencing its effects? Suddenly the square-like shapes and triangle-like shapes aren’t looking so different are they? And have you noticed how all the round shapes seem sort of ‘like’ one another? Our shapes are going from different to similar without changing their features (only the frame of reference has changed).

Remember when we said that “really there are no such intrinsic similarities or differences”. This is what we were talking about. Similarity isn’t a stable characteristic of our sensory stimuli. Instead, similarity is something that we add to the stimuli through categorization. Said otherwise, similarity doesn’t drive categorization; categorization drives similarity.

Did I just blow your mind?

Probably not. But I still want to let the above sink in. In some ways comparative fit is one of the biggest contributors of impact to self-categorization theory, and consequently the social identity approach. As such, we shouldn’t move past this too quickly. To sum up succinctly then…

The principle of comparative fit is that we are more likely to believe that a collection of stimuli represents a group or category to the degree that the observed distances between those stimuli are on average less than the observed distances between that collection of stimuli and other stimuli in the frame of reference.

Expect posts about normative fit and perceiver readiness to come soon. There is more to say, but once we have these we will have a pretty comprehensive picture of the way in which we carve up our social world and find our place in it.

James Clifford, CC-BY-SA 3.0 via Creative Commons

Where do you see your place?



[1] Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group behavior. Advances in group processes: Theory and research, 2, 77-122.

[2] Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the Social Group: A Self-Categorization Theory. Blackwell: Oxford.

[3] Usually comparative fit is described in terms of observed “differences” rather than “distances”. However, we take to heart the argument from Penny Oakes, Alex Haslam, and John Turner that the term ‘differences’ might sound like a circularity.[5] The term they suggest in full is “precognized distances”, where ‘precognized’ makes explicit the fact that we are talking about raw, not yet processed, sensory input. We stick with just ‘distances’, but emphasise this idea in similar language a little further on in this post.

[4] McGarty, C. (1999). Categorization in social psychology: Sage Publications Ltd, p 112.

[5] Oakes, P. J., Haslam, A. S., & Turner, J. C. (1994). Stereotyping and Social Reality. Oxford: Blackwell.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s