Accounting for Divorce in Marital Research: An Application to Growth Mixture Modeling

• Author: Jeremy B. Kanter,J. Kale Monk,Christine M. Proulx
• Date: 01 April 2019
• Published date: 01 April 2019
• DOI: http://doi.org/10.1111/jomf.12547

J B. K , C M. P,  J. K M University of Missouri

Accounting for Divorce in Marital Research: An

Application to Growth Mixture Modeling

Objective: The goal of this study was to under-

stand the implications of omitting versus

retaining individuals known to eventually

divorce in the longitudinal modeling of marital

quality trajectories.

Background: Change in marital quality has

been the focus of basic and applied research

as well as policy initiatives for the past sev-

eral decades. Scholars have used group-based

modeling techniques to better understand how

enduring marriages develop over time. Previ-

ous studies, however, have not directly exam-

ined the implications of including versus exclud-

ing individuals who go on to divorce for shape

or number of identied marital development

trajectories.

Method: The present study used growth mix-

ture modeling and the Marital Instability over

the Life Course data set to examine patterns

of marital happiness. Analyses were rst con-

ducted only on continuously married men

(n=269) and women (n=438). Identical ana-

lytic procedures were then used with combined

samples of those who were continuously mar-

ried plus those who eventually divorced (n=358

men; n=588 women).

Results: The results indicated that retain-

ing those who go on to divorce increased

the number of classes reported and changed the

Human Development and Family Science, University of

Missouri, 314 Gentry Hall, Columbia, MO 65211

(jeremykanter@mizzou.edu).

Key Words: divorce, longitudinal research, marital quality,

methodologies.

composition of class membership and patterning

of marital quality over time for both men and

women.

Conclusion: The study concludes with a recom-

mended three-step procedure for modeling mar-

ital change when samples include individuals

who divorce following a baseline assessment.

The study of marital development and change

remains a popular interdisciplinary topic of

research (e.g., Anderson, VanRyzin, & Doherty,

2010). Recently, group-based modeling has

become a common analytical technique to study

how marriages develop and change over time

(Proulx, Ermer, & Kanter, 2017). Group-based

modeling identies distinct groups of indi-

viduals within a sample who follow similar

developmental trajectories (Wickrama, Lee,

O’Neal, & Lorenz, 2016). Despite the method’s

popularity for understanding various pathways

of marital development, few scholars have

explored sample decisions researchers must

make, such as the implications of including

versus excluding individuals who go on to

divorce in the analysis. Although some studies

have intentionally included individuals who

divorce prior to the study’s end to examine

the association between change in marital

quality and divorce proneness, it is unclear if

this inclusion affects the characteristics of the

trajectories reported (e.g., patterning or number

of classes). Sample size alone should not inu-

ence the number of classes identied beyond a

threshold of 300 to 500 respondents; however,

sample inclusion criteria, such as divorce status,

might inuence the numbe of classes, patterns

520 Journal of Marriage and Family 81 (April 2019): 520–530

DOI:10.1111/jomf.12547