An important argument of the social diversification hypothesis is that disadvantaged groups employ the Internet rather than face-to-face communication to widen social networks, whereas advantaged groups employ the Internet to strengthen extant network ties. Prior inquiry in this area has not accounted for both online and off-line communication, has only been explored with cross-sectional data, and has mainly been studied in Israel. To deal with these gaps with a U.S. data set, 2,669 conversations were examined over 6-day periods employing ecological momentary assessment (EMA). Unlike participants from racially or educationally advantaged groups, participants from a racially marginalized group or lacking college training were more likely to widen social networks online rather than face-to-face with interracial and weak tie exchanges. This proof of concept of social diversification theory across cultures is the first to employ real-time, within-person measures of both race and tie strength.