This research is a test and extension of Tichenor, Donohue and Olien's knowledge gap paradigm (1970) which holds that as information diffused in a social system increases, gaps in knowledge will tend to widen between lower and higher socioeconomic strata of the population. Investigating motivation-based gaps, the study extends beyond extant research by examining the role of more complex interactions in producing gap effects. The context of the study is a Medicare/Health Maintenance Organization multi-channel informational campaign to promote knowledge (and adoption) of the HMO delivery system among elderly. Expanding the tradition of gap research on special populations, the study examines one whole population substratum, the elderly. It is one of the first investigations of HMO information efforts to the general elderly population, rather than to older HMO members. The aged's disproportionate utilization of health care resources suggests that health delivery systems are salient to many elderly. The hypothesis that differential salience will cause gaps in knowledge is investigated. Two salience indicators are used: an individual's health status and supplemental health insurance status. Three main research questions are: 1) Does the classic knowledge gap hold for these data? 2) Does the salience knowledge gap hold? 3) Is the knowledge gap effect the result of some more complex interaction between salience and education? Using an original instrument, 401 Medicare beneficiaries were interviewed in person: 109 had received substantial communication intervention, 292 received minimal information. In addition to measures of education and salience, the instrument included measures of HMO knowledge and demographic variables used as controls (age, marital status, race, and occupational prestige). The analysis uses means comparisons (t-tests) for illustrative purposes, and tests hypotheses and interaction effects with multiple regression analyses. The findings were: the communication intervention sharply affected knowledge. The classic education knowledge gap was supported, but the simple salience knowledge gap was not. However, the most interesting result is that the education knowledge gap advantage held only among the ill, and not among the well respondents. Thus, the essential finding is that simple side-by-side comparisons in gap research stop short of revealing the specification effect shown here in the interaction between socioeconomic and salience indicators.