The paper analyzes the degree to which subjects are able to "accurately" report the attitudes of their reference group or close discussion partners. Studies using a social network approach have found evidence for a strong association between the respondents' own attitude reports and their reports on the perceived attitudes of relevant others. Since most research has not tested whether this perception of ego is congruent with the actual attitudes of the alters, it remains unclear to what degree the observed correlations are substantiated in social influence processes or simply result from egos' projection of their own attitudes on their alters. However, descriptive statistics have confirmed that proxy-reports and self-reports with regard to attitudes reveal a relatively low degree of congruence. In this study, the influence of ego's access to information on the "quality" of the proxy-reports concerning attitudes towards immigrants in Germany is analyzed in the context of a broader understanding of the congruence of proxy-reports. For this purpose, we use two approaches that apply different criteria of validity to assess the "accuracy" of proxy-reports on attitudes on an individual level. In order to address our questions, we reanalyze recently collected data (N = 1693) on egos' attitudes towards immigrants, including information about egos' perception of alters' attitudes and alters' self-reports on the same matter. According to our results stemming from multivariate two-level regression models, the accuracy of proxy-reports is conditional on characteristics of the dyadic relationships as well as the complete network in which these dyads are embedded, a fact which apparently has not yet been considered in previous research. In summary, it is recommended that for the collection of attitudinal data by means of proxy-reports the name interpreter items should refer to general dispositions rather than to specific attitudes of alters.