Models of Non-Independence in Dyadic Research

TLDR

Dyadic research often involves non‑independent responses between partners. The study presents statistical estimation methods for three processes that generate non‑independence in dyads. The authors model three mechanisms: cross‑partner influence of X on Y, reciprocal influence of Y variables, and joint X‑Y scores reflecting couple‑level factors.

Abstract

In dyadic research, the responses of the two members of the dyad are likely to be non-independent. Statistical estimation for three different processes that bring about non-independence are presented. In each, a variable X influences Y. First, the X variable of one person may influence the Y variable of his or her partner. Second, the Y variables of the two people may directly influence each other. Third, the two X and Y scores may be indicators of couple-level factors.