An Entitativity Measure and Why

For all you folks out there google searching for Entitativity (and there may be a few) and, in particular, for those of you looking for a validated measure of entitativity, I’d like to direct you to our published peer-reviewed paper in Group Processes and Intergroup Relations.

For the moment, though, let me share why my lit search this morning on the latest entitativity publications has gotten me all worked up. Grab yourself some popcorn and read on. All of the following info is in our article, so read this post. But please read and cite from our article.

{rant} C’mon folks!! Think a little bit about how you measure what you study and even more importantly how you define it. Please read Suddaby’s editorial on how essential construct clarity is while you are at it.

First, if you say your construct (let’s say, oh, entitativity) is related to another construct (let’s say, oh, cohesion) then you need to be extra sure that any measure of entitativity DOES NOT CONTAIN ANY PART OF THE MEASURE OF COHESION!!! PEOPLE! That makes me type in all caps! If you believe entitativity and cohesion are two different constructs (ahem, as in Study 2 of our article) BUT you use a measure of entitativity that CONTAINS cohesion questions, then one of two things will happen**. You will run a CFA and determine that, by golly, that entitativity item inappropriately loads onto the cohesion scale, and you should eliminate it. Or (and this is what I have just seen multiple examples of!!), go ahead and collapse the items onto their scales (thrilled that your Cronbach alphas as so high!!!) and run correlations, regressions, and ANOVA analyses showing strong relationships. But, statistically, you are cheating. You are over-estimating the relationship because that cohesion item that you’ve included in your entitativity scale wants to be with its friends on the cohesion scale; it correlates with its friends on the other scale. Your results, really, should not be trusted. Your results come from an overlap of the items not a pure relationship between the constructs. There’s a similar problem with affective commitment which contains measures of identity and retention. No wonder affective commitment correlates so highly with retention and identification! ((I use this measure of Organizational Commitment now. It’s a cleaner measure of the construct))

Second, if you say that “entitativity is composed of these particular characteristics of a group”, take a minute and unpack that argument. First, I can’t even write entitativity is the characteristics of the group. That equates entitativity with group characteristics. Entitativity is a perception. Behold, are these people waiting at a bus stop a group?

People at Bus Stop

Well, they are a “group” of people waiting at a bus stop. But what about them now??

People in Lab Meeting

It’s the same group of people, but they DEFINITELY seem more like a group here, right? They are interacting, sitting around a table, working on something together. They are definitely a groupier group in the lab than at the bus stop!! That variance in your perception about how groupy they are is called entitativity. And social units vary from low entitativity (at the bus stop) to high entitativity (in a lab meeting). Entitativity is a person’s cognitive assessment of a social unit as a group (Blanchard, Caudill, & Walker, 2018).

But what changed in the pictures above? The characteristics of the group, right!!?? Because they are interacting, sitting around a table, working on something together they look more like a group. The characteristics of this group caused us to assess them as more of a group.

Folks. Whenever you use the words “cause,” “because,” or “why” you are looking at an explanation for a relationship you have observed. If A causes B, then A is not a characteristic of B, it is a FREAKING cause. Or, in this case, an antecedent. These group characteristics (interactivity, similarity on working towards a goal, physical boundaries) are antecedents to entitativity. DO NOT INCLUDE THE ANTECEDENTS OF A CONSTRUCT IN ITS MEASURE!! Do not just measure the antecedents and say you have the construct, for Pete’s sake. Measure the antecedents! Measure the construct! Test the relationship! Research should be about the truth. Measuring the antecedents/characteristics and saying you measured the construct is not the truth. Make sure you have truth in your definitions and in your measures.

You are messing up the theory as well as the measure if you do not have a clear, concise definition of your constructs as well as clear, validated operationalizations of your measures. Don’t believe me? Use a confirmatory factor analysis on your data. (NOT an exploratory factor analysis) That sh!t will not let you lie.

And don’t even get me started on trying to equate gender and a softball team as mostly equivalent types of groups. Let me take a break before I get cranky about that theoretical and empirical fallacy.

**This is what caused my head to explode this morning. Seeing someone use cohesion in a measure of entitativity.

About Anita Blanchard

Associate Professor, Department of Psychology
This entry was posted in News. Bookmark the permalink.