The safety training programme here can be considered to have internal validity in so far as it has achieved attitude change for the higher grades (Goldstein, 1991). It was conducted participatively in small groups, which would seem to be an effective approach (Nyandindi et al, 1995; Chapman et at, 1996; Simon et al, 1996). However, the safety programme was initiated by management: this was inevitable since there was considerable cost involved. Further, the trainers were themselves "appointed" by management before being trained by the consultants. Thus, there may be issues of perceived "ownership" by management who would, after the training programme had taken place, be likely to demonstrate more positive attitudes and behaviour in relation to safety to avoid cognitive dissonance (Festinger, 1957). The shop floor who may identify less with the programme experience little or no dissonance and were therefore less likely to change their attitudes. Dissonance theory is clearly one possible way of accounting for the findings in this study. The results focus of the training in terms of action outcomes, not generated directly by the teams themselves, may also have contributed to these differences in responses and may contribute to teams' lack of empowerment and even to a loss of agreement about the goals of the training (Meyer, 1994).