Research design We test the proposed hypotheses using a 2 × 2 between-subject controlled experiment. The two independentvariables are type of incentive scheme (formula-basedscheme versus subjective weighting scheme) and whetherthe strategic value of HIA is communicated to employ-ees using a strategy map (presence versus absence of communication). The dependent variables are participants’reported willingness to share knowledge with a co-workerand to perform extra-role behaviours in a fictitious consul-ting firm (discussed in Section 3.2).3.2. Research task and dependent variablesParticipants assumed the role of a management con-sultant at a consulting firm, Alpha Consulting. Participantswere told that the overall goal of Alpha Consulting was to increase fee revenue and, in line with this goal, theirmonthly performance would be evaluated based on four performance areas: billable hours, billable rates, dollarvalue of future work requested by clients, and client’sevaluations.4Using a scenario similar to Chow et al. (2000),participants were told that, over the last few years, they had built up substantial knowledge and expertise about Industry X, such as a list of useful contacts and relevantbenchmark information. To measure the first dependent variable, willingness to share knowledge, participants wereasked whether they were willing to spend their time shar-ing this knowledge with a new colleague, Jones, in anotherdepartment (Tax Advisory Department) of Alpha Consul-ting, who was given a portfolio of tax clients in Industry X.It was made clear that although Jones came from a differ-ent department (Tax Advisory), and therefore was not indirect competition with the role assigned to participants,the amount of time spent on sharing this knowledge would result in a delay to one of the client projects the participantsaimed to start this month and would also lower the num-ber of billable hours for the participant in this month. Ourscenario therefore suggests potential costs to knowledgesharing.To measure our second dependent variable, general ten-dency to perform extra-role behaviours, participants werealso asked to respond to six questions based on Williamsand Anderson’s (1991) OCB-I scale. Specifically, these ques-tions ask participants to indicate the extent to which theyare willing to help co-workers when they have been absent,help co-workers with heavy workload, take time to listen to co-workers’ problems, go out of their way to help new co-workers, take a personal interest in co-workers, and pass along information to co-workers.