Premium
Investigating Family Shared Realities With Factor Mixture Modeling
Author(s) -
Dyer W. Justin,
Day Randal D.
Publication year - 2015
Publication title -
journal of marriage and family
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.578
H-Index - 159
eISSN - 1741-3737
pISSN - 0022-2445
DOI - 10.1111/jomf.12158
Subject(s) - unobservable , psychology , factor (programming language) , class (philosophy) , mixture model , social psychology , developmental psychology , set (abstract data type) , structural equation modeling , sample (material) , econometrics , mathematics , computer science , statistics , artificial intelligence , chemistry , chromatography , programming language
The authors introduce factor mixture modeling ( FMM ) and its implications for assessing varying degrees of shared reality in families. FMM extends the typical factor analytic model that uses a set of observed indicators to identify an unobservable factor and its structure. FMM examines the possibility a sample is a mixture of “classes,” each with unique factor structures. To demonstrate, varying factor structures of father involvement are identified using father, mother, and child reports as observed indicators. Three classes were identified and relate to varying degrees of shared reality regarding the father. Predictors and outcomes of these classes are modeled. The class in which the child's view most overlapped with his or her parents' view was predicted by higher child agreeableness and higher levels of maternal depressive symptoms. As hypothesized, membership in this class predicted lower levels of parental conflict and fewer parental attempts to involve their children in their conflicts.