Hierarchical Multicomponent Model of In-Group Identification

What it is

The paper opens from a problem of agreement. In-group identification is psychologically important and socially consequential, the authors note, but researchers had little agreement on how to conceptualise or measure it. Their answer is a model with two levels.

Under self-definition sit two components. Individual self-stereotyping is seeing oneself as similar to the typical member of the group. In-group homogeneity is seeing the members as alike and distinct from outsiders.

Under self-investment sit three. Solidarity is attachment and connection to the group. Satisfaction is a favourable judgement of the group. Centrality is how prominent membership is in a person's life.

The model sits between two other positions in the literature on group identification. Naomi Ellemers and colleagues had argued in 1999 for three separate aspects, and Tom Postmes and colleagues proposed in 2013 that one item can capture social identification. Leach's five components give more resolution than the first and ask more of a questionnaire than the second.

In effect

Seven studies tested the model. The first two used confirmatory factor analysis across three different group identities to check that the five components hold together in the hierarchical structure. The next two compared the components with established identification measures, which tests construct validity. The last three asked whether the components prospectively predicted how people orient to, and feel about, real relations between their own group and others.

The authors concluded that the hierarchical multicomponent model has conceptual and empirical value. In our opinion the practical use is diagnostic: when overall identification changes, the components can show whether the change is in how people see themselves or in how much they invest in the group.

What it does not say

The model does not say that every group needs all five components measured, and it does not rank the components by importance. It does not claim to replace shorter measures, and the single-item approach published five years later takes a different view of how much a short measure can carry.

The page rests on the paper's own summary of its aims, studies and conclusion. It reports no effect sizes, because none are needed to describe the structure, and it does not extend the findings beyond the group identities the authors tested.

---