A New Asymmetric Measure of Association for Ordinal Variables

table of means (Table 3) suggests that the influence exerted on topics by the less educated opinion leaders may be colored by ideological orientations different from those among the better educated respondents. It would be of further interest to learn whether opinion leaders' sense of conservatism leads to differential selection of topics about which to exert influence. For instance, one might hypothesize that (controlling for education) the highly resistant opinion leaders confine themselves more to local issues, and the more liberal leaders exert influence on a more cosmopolitan range of topics. Similarly, conservative and liberal opinion leaders might be found to differ on level of information about topics (again, controlling for education). It is somewhat disquieting to learn that for this sample high opinion leaders are as likely as the non-leaders to agree with statements expressing an anomic reaction to life, such as that there's little use writing officials, because often they aren't really interested the problems of the average man, that nowadays, a person has to live pretty much for today and let tomorrow take care of itself and that in spite of what some people say, the lot of the average man is getting worse. This finding leads one to speculate (hopefully) that perhaps the twoitem opinion leadership typology is too permissive or that the term public affairs is not a meaningful cue to respondents. The interaction noted between age and sex on opinion leadership implies that leadership the area is linked to respondents' life cycle. It may be that younger men actively pursuing careers and upwardly striving find that knowledge and influence concerning politics are instrumental and rewarding, while at the same time their wives are devoting energies to the domestic duties of raising families. Later life, when, freed from home responsibilities, the women may find increasing opportunities and social rewards attendant upon opinion leadership. Finally, it is not surprising that opinion leadership is substantially correlated with education (.47) and with self-ascribed social class (.41).