Premium
CATEGORIES OR DIMENSIONS: LESSONS LEARNED FROM A TAXOMETRIC ANALYSIS OF ADULT ATTACHMENT INTERVIEW DATA
Author(s) -
Ruscio John
Publication year - 2014
Publication title -
monographs of the society for research in child development
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.618
H-Index - 63
eISSN - 1540-5834
pISSN - 0037-976X
DOI - 10.1111/mono.12119
Subject(s) - spurious relationship , psychology , categorical variable , harm , dimension (graph theory) , social psychology , representation (politics) , preference , cognitive psychology , computer science , statistics , mathematics , machine learning , politics , political science , pure mathematics , law
Booth‐LaForce and Roisman's monograph on the Adult Attachment Interview (AAI) featured a taxometric analysis to determine whether variation along two components, dismissing and preoccupied states of mind, was categorical or dimensional. Empirically evaluating the latent structure of these constructs helps to avoid spurious categories or dimensions. This benefits researchers working with measures of adult attachment to maintain as much predictive validity and statistical power as possible, and it benefits researchers who build or test theories of adult attachment by steering the search for causal factors in fruitful directions. Fraley and Roisman (Chapter 3, this volume) performed their taxometric performed their taxometric analyses in an exemplary fashion, adhering carefully to empirically supported, practical guidelines. They adopted an appropriate inferential framework for their taxometric results that pits two competing structural models against one another. They were willing to accept that the taxometric results for preoccupied states of mind were ambiguous and they tentatively advocated a dimensional measure on the grounds that, even if this was not the best representation, using a spurious dimension might do less harm than using spurious categories. Rather than embracing a general preference for categories or for dimensions, researchers should evaluate the pros and cons of each potential structure‐measurement mismatch on a case‐by‐case basis.