producing evidence that some teams generated stress and work intensification. Debates on teams, and also on the closely related issue of total quality management, often took on an essentialist character, with proponents singing their praises and critics exploring their negative features (Wilkinson et al., 1997). Further research refined the analysis, in three main respects (Edwards et al., 2002).
First, rather than assuming that there was a uniform phenomenon called teamwork, analysts identified different kinds of teams. A major distinction was between the teams characteristic of lean production and those based on ideas of autonomous work groups. In the former case, team autonomy was tightly restricted whereas in the latter it was much more extensive. The causal power of ‘teams’ thus varied according to the type of team involved. Other more detailed distinctions were also introduced, for example the degree to which managements displayed real commitment to team ideas. Such commitment was shown to be important in shaping the extent to which the potential for teams to deliver desired outcomes was realized in practice.
The second refinement was to address the context in which teams operated. The ‘teamwork dimensions’ model identified three aspects of the context, which were labelled the normative, governance, and technical dimensions (Thompson and Wallace, 1996). The point was to show that the ‘effects’ of teams were shaped by each of these dimensions, rather than existing independently. Empirical research in a single company in different countries showed that the ‘same’ teamwork ideas worked quite differently in different national contexts (Thompson et al., 1995). To take another example, Ortiz (1998) used case study methods to examine teamwork in General Motors in Spain. He framed the analysis in terms of largely critical views of teams from a union and worker viewpoint in the US and UK and more favourable responses in Germany and Sweden. He showed that Spain fell between these two extremes, which he explained in terms of the way in which the IR system allowed negative features to be contained and positive features to be developed. Hence, in the UK, teams are often seen as a threat to trade unions, because they can undermine unions’ claims to be the sole legitimate voice of workers; teams set up alternative channels, and in the absence of strong legal support for unions the danger is that firms will decide to operate without unions. In Spain, by contrast, unions were institutionally entrenched, and teams were not so much of a threat. In CR terms, teamwork has causal powers of positive and negative kinds, and how they are actualized depends on specific conditions.
Third, researchers identified sets of conditions that made teamwork more or less successful. These conditions included the extent of job security and the structure of workplace relations into which team experiments were introduced. It should be stressed at this point that a factor such as job security is not to be seen as having necessary effects. It is true that studies have shown that an absence of security tends to undermine team experiments, because co-operation and worker commitment through teams is unlikely where the employer’s commitment to maintaining jobs is weak. As Thompson (2003) has argued, there are important structural features of modern capitalism that make it hard for employers to keep their promises (even when these are genuine). Yet other studies have revealed cases where job security has been weak but where a degree of success of teams was achieved. One set of reasons concerns employee expectations: where there has been little history of employee involvement, even quite modest moves can be valued even though security remains uncertain (Marchington et al., 1994). A second set of reasons concerns the context of a team experiment. Where workers lack alternative jobs, and where they know that an