The TOP 12 is a test designed to assess collective (semantic) memory quickly and easily. It relies on eight questions about the life of 12 celebrities that are named (verbal presentation). Test validation is often based on the definition of a criterion external to the test (i.e. a pathological threshold, external validation); internal validation is itself rarely studied. In the present study, our aim is to assess the internal validation of the TOP 12.

To show how the responses provided by the subjects measure a single hypothetical dimension (also called a construct): collective memory ability.

The sample was made of 145 subjects (91 control subjects, 32 patients with Alzheimer's disease, 21 with amnestic mild cognitive impairment (aMCI) and 1 with semantic dementia). Two stages of analysis were required: modelling the responses for each item using a dedicated model with three parameters and testing for the unidimensionnality of the estimated scores.

The eight models fit tightly to the data. A combinatorial factorial analysis does not allow rejecting the idea that the eight questions assess one and only one hypothetical dimension.

Psychometric modelling of the TOP 12 data shows that they assess collective (semantic) memory.