The study of loneliness and social isolation has provided a lot of evidence for differences in the prevalence of the two, depending on the context of individuals. Given different social, legal, and economic differences for migrants and refugees, it has been documented that these groups show elevated levels of both social isolation and loneliness compared to the respective host population. Differences in association between social isolation and loneliness have received less emphasize. We test five competing hypotheses about the different sizes of association between social isolation and loneliness in the groups of migrants, refugees, and the host population in Germany. The hypotheses are informed by the differences in social, legal, and economic circumstances between the groups and their socioeconomic and psychological consequences. Using survey data from a large stratified random sample of the population, including migrants and refugees, we test our five hypotheses using a Bayesian Evaluation of Informative Hypotheses. We find highest relative support for the hypothesis about increased need for social networks and support among refugees, which would be indicated by a higher association of social isolation and loneliness for refugees than for the host and migrant population. However, further investigation of the results show all theory derived hypotheses perform poorly in explaining the major pattern in the data: The association of social isolation and loneliness is lowest for migrants (about 0.25 SD), with similar larger associations for refugees and the host population (about 0.5 SD). We discuss this contradiction of theory and evidence, proposing avenues for future research.