Longitudinal designs and constrained-choice settings best establish contact effects
Evidence
Find an intergroup situation that severely limits choice to participate. In the Merchant Marine, police, and public-housing research, little choice was available to participants. … 3. Although not without problems (Clogg 1986), longitudinal designs are best (Pettigrew 1996). Yet such research is rare in intergroup research; Sherif’s (1966) study is a famous exception.
— Thomas F. Pettigrew, Intergroup Contact Theory, 29
Categorization
XREF: Mirrors study-design hierarchies used in epidemiology and program evaluation, where natural experiments and panel data beat cross-sections.