A study examined mass media use and ability to predict public support for two heavily publicized Florida referenda--casino gambling and a statewide lottery. It was hypothesized that (1) voters who read newspapers that publish election public opinion poll stories will be better able to predict support for the issues than voters who rely on friends and family for election information, and (2) voters who rely on mass media (other than newspapers) for election information will be less able to predict support for the issues than voters who rely on newspapers but more able to do so than those who rely on friends or family. Subjects, 666 individuals from all over the state, were surveyed by telephone two weeks before the November 4, 1986, election. Neither hypothesis was supported by the data collected. Results indicated that it was impossible to say why some respondents more accurately predicted support for the two referenda, but that some evidence existed implying that patterns of media use were not a factor. Examination of the data did not reveal whether respondents read or watched poll stories, whether they saw them and forgot them, whether they ignored them, or whether they simply did not believe them. Results may also suggest to editors that reader interest may not be sufficient justification for conducting and reporting public opinion polls. (Tables of data and references are appended.) (NKA)