{"id":4642,"date":"2025-01-18T11:59:48","date_gmt":"2025-01-18T10:59:48","guid":{"rendered":"https:\/\/www.psyctc.org\/psyctc\/?post_type=docs&p=4642"},"modified":"2025-01-20T17:08:19","modified_gmt":"2025-01-20T16:08:19","password":"","slug":"bifactor-models","status":"publish","type":"docs","link":"https:\/\/www.psyctc.org\/psyctc\/glossary2\/bifactor-models\/","title":{"rendered":"Bifactor models"},"content":{"rendered":"\n

A subset of factor analytic models, not to be confused with 2-factor models.<\/p>\n\n\n\n

Details<\/h4>\n\n\n\n

As with all factor analytic analyses bifactor models are a set of models of how we might understand correlations between scores on multiple variables (scores collected across a lot of different peopleand such that scores from any one person are not related to those from another person, e.g. not collected by collecting all people in one house, a house at a time). Bifactor models are a particular subset of confirmatory factor models. The bifactor model assumes that the scores on all the variables come from one of three sources:
1) a general factor
2) “group” factors each involving only some of the variables (and no variables contributing to more than one group factor)
3) “specific” or “error” contributions to each variable, uncorrelated with either the of the above factors. I.e. “error”, “noise” or “unreliability” variance.<\/p>\n\n\n\n

These models are widely used where it is assumed that there is some general factor so a popular model of psychological distress or dysfunction is that we each have a general tendency to any distress, a general or p<\/em> factor, but that we can also have distinct amounts of more specific forms of distress\/dysfunction, often diagnostic groups like anxiety or depression.<\/p>\n\n\n\n

These models can <\/em>have some real strengths but are often used rather uncritically. The pros and cons are really beyond this glossary but I might get into them in the Rblog and even, at some point, in a paper.<\/p>\n\n\n\n

Try also<\/h4>\n\n\n\n