Do personality test responses actually cluster into distinct groups?

Psychologists typically use Latent Class Analysis to find hidden groups in survey responses, but it requires guessing the number of groups upfront and struggles when questions are correlated. This team built a network approach: treat each person as a node, link them by similarity, then search for natural communities. Testing 14 standard personality scales, they found most produce artificial subgroups driven by the questionnaire's structure itself. Only a handful show genuine modular patterns independent of these artifacts.