Overview
This series asks what evidence can actually show when intergroup contact and outgroup attitudes move together. It separates between-person association, within-person longitudinal change, and intervention-relative effects, then examines how CLPM, FF-CLPM, and RI-CLPM allocate variation across those questions. Across 31 datasets from 21 publications and 33,348 participants, stronger control leaves the general naturalistic within-person effect small and uncertain, while stable between-person associations remain substantial. A field study of Soliya’s Connect Program found encouraging changes across warmth, self–outgroup overlap, dehumanization, and meta-dehumanization under repeated, facilitated, collaborative virtual contact. Its non-randomized comparison, selective attrition, bundled supports, student sample, and mixed sensitivity results limit definitive causal claims.
Learning outcomes
- A between-person contact–attitude association does not establish that increasing contact will change an individual’s attitude.
- The re-analysis found substantial stable associations but a smaller, uncertain within-person signal under stronger-controlled models.
- Soliya’s repeated, facilitated, collaborative virtual program was associated with favorable changes across four measured dimensions in its primary analysis.
- Non-randomized assignment, selective attrition, bundled supports, and mixed sensitivity results limit definitive causal claims.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
What is the difference between a contact association and a contact effect?
A contact association compares people who differ in contact, whereas a contact effect asks whether changing contact changes a person’s later attitudes. The first does not by itself identify the second.
Why can a longitudinal contact study still fail to show causation?
A longitudinal contact study can fail to show causation because stable or time-varying factors may influence both contact and attitudes despite temporal ordering. A lagged path is not a clean counterfactual without stronger confounding control.
What did the 31-dataset re-analysis find?
The 31-dataset re-analysis found substantial stable between-person associations but little statistically secure general naturalistic within-person signal under stronger-controlled models. Low power, varied timing, and heterogeneous measures leave small context-dependent changes possible.
Does virtual intergroup contact improve attitudes?
Soliya’s structured virtual program was associated with improvements in attitudes, self–outgroup overlap, dehumanization, and meta-dehumanization in its primary field analysis, but its non-randomized design and mixed sensitivity results are not randomized proof.