A study examined how use of different kinds of "involvement" measures (behavioral, cognitive, or affective involvement, gratifications sought, and cognitive effort expended) could help explain differences between people's knowledge of election information and whether the way that knowledge is measured affects conclusions about involvement with television and newspapers. Respondents, 238 adults surveyed by telephone, completed a questionnaire covering media use and attitudes, perception and knowledge of candidates and issues, voting patterns, political beliefs, and demographics. Results indicated that, in addition to observed differences in information holding usually explained by group differences in political interest or education, three other explanatory factors contributed to observed differences: (1) the operational definition researchers used in measuring knowledge, (2) the amount of cognitive effort expended in processing information from newspapers and television, and (3) respondent's age. Results also indicated no difference between newspaper- and television-reliant individuals' recognition of election information, but did indicate greater recall of election facts among younger, newspaper-reliant respondents, suggesting that television-reliant individuals are receiving and processing election information, but that the information can best be retrieved through a multiple-choice type of question. Findings show that when recall is measured, newspaper-reliant people may appear more "involved" in the election, but when recognition measures are used, television-reliant people do just as well. Thus, conclusions about whether a medium is "high" or "low" involvement may depend totally on the subjects. (Extensive tables of data, figures, notes and references, and the study instrument are appended.) (NKA)