After reviewing the literature on item non-response we focus on three issues: First, is there significant heterogeneity in item non-response across financial questions and in the association of covariates with item non-response across outcomes? Second, can the informational value of surveys be improved by matching interviewers and respondents based on their characteristics? Third, how does offering a "don't know" answer option affect respondent behavior? The questions are answered based on detailed survey and interviewer data from the German Socioeconomic Panel, considering a broad set of income and wealth outcomes.