Previous studies of immigrant populations suggest that ceteris paribus (after controlling for the number of years in the receiving country and other socio-demographic variables), the level of income is strongly and positively correlated with fluency in the local language. Based on a phone survey held in 2005 among a representative sample of Former Soviet Union (FSU) immigrants, the current study extends this literature and investigates the possibility that the standard model is misspecified. Unlike previous surveys, our dataset includes detailed subjective questions on the degree of social involvement. Our findings indeed support the conclusion that the standard model is misspecified. At 1% significance level, immigrants who are better assimilated within the receiving country are 11% more likely to attain a level of income that is equal to or higher than the average level of net family monthly income. Moreover, compared to the incorrectly-specified model, at 1% significance level a shift from lower to intermediate and high level of language proficiency does not significantly increase the level of income. Consequently, marginal probabilities of income shift, which have been mistakenly attributed to better language proficiency in the misspecified model, should have been, in fact, attributed to a higher level of social involvement. Finally, stratification of the sample based on gender and marital status shows that compared to unmarried females, married males have a higher return on social involvement. Among married men (unmarried women) a higher level of social involvement significantly increases the chances for higher income level by 15% (only 4%). Research findings thus stress the important role of better social involvement, particularly among married males: a higher degree of social involvement leads to improved social networking and, in turn, to better job opportunities and higher income.