As discussed elsewhere (Neil and Snizek 1987), the decision as to whether to use structural or perceptual criteria in the measurement of discrimi- nation presents a dilemma. On the one hand, structural criteria used in studies that adopt a human capital approach to gender inequality fail to measure more diffuse forms of discrimination, such as the use of differen- tial methods of workplace control for men and women (Acker and van Houten 1974), differences in the isolation and interpersonal attractive- ness of men and women in the work context (Miller et al. 1975), lower visibility and lack of integration of women in the scientific community (Lonsdale 1970; de Meurion-Landalt 1975; Stolte-Heiskanen 1982), and gender-linked differences in the assessment of male and female scientists' professional reputations (Tuchman 1980). These forms of differential treatment can, in turn, influence productivity, citations and aspirations and thus disguise differential treatment of men and women as non-dis- criminatory, on the basis that it is a response to differential behaviour. By
contrast, it-has often been noted that a discrepancy may exist between the perception of discrimination and its actual occurrence (Morlock 1973: 293; Cass et al. 1983: 94 ff; Hoffman and Reed 1981; Crosby 1982).
In this study, it proved impossible to get structural indicators for many of the forms of discrimination likely to occur - although analysis of com- pany personnel statistics (Subcommittee on the Employment of Women 1983) showed that women were underpromoted in both the research and administrative career streams. As noted in the Subcommittee's Report (1983: 12), in the junior research stream, 'there is a tendency for women to be preferentially assigned to work which is repetitive and/or routine,
offers little hope for promotion, and is less securely funded'. Accord- ingly, a decision was made to use perceptual measures of job-related gender discrimination, as these pertained to both day-to-day interaction and promotion.
Two sets of items were used to measure women respondents' perceptions of job-related discrimination. The first set measured inter alia, perceived discrimination on a day-to-day basis. Here respondents were presented with a number of work situations and asked whether, in the previous three years, they had experienced favourable, irritating or very irritating