Are there distinct cognitive types?

A few previous attempts to isolate cognitive types-groups of individuals with similar profiles of cognitive skills-have been reported in the literature. For the most part, they have not arrived at the same types. The present study sought to identify cognitive types via cluster analysis in an available sample of adolescent twins. Analyses were carried out using one randomly selected twin from each of 16-year-old Australian twin pairs who had taken a five-scale cognitive test battery (N = 677 pairs-a further sample of 628 pairs was reserved for replication purposes). Three levels of tightness of clusters were examined-minimum mutual intercorrelations among cluster members of 0.10, 0.30 and 0.50. The higher intercorrelations led to more clusters-18 at the 0.10 criterion, 25 at 0.30, and 36 at 0.50. The 0.10 level was most similar to the number of clusters reported in previous studies, and was chosen for further analysis. Clustering at this level yielded 18 clusters of individuals from the random twins, and 17 from their co-twins. Fifteen of these matched, as evidenced by a correlation of 0.80 or more between the cluster means. A second method of clustering based on a different approach, SPSS Cluster, gave similar results-15 clusters matched those from the original analysis. Agreement for cluster membership was compared for identical and fraternal twins. A greater agreement for identical twin pairs was found, and interpreted as evidence of a genetic contribution to the clustering. Most of the above analyses were successfully replicated on the reserved sample of twin pairs. Thus there was evidence for stability of initial clustering in this population. However there was not much evidence across studies for discrete and dependable cognitive types.