CHAPTER 17

Research Methods

To test our hypotheses, we gathered survey data from 94 project teams. All projects included in the sample were at least 50% complete or had been completed within the past 6 months. This requirement ensured that the team’s work was: (1) sufficiently underway to enable accurate assessment of it’s functioning or, (2) recent enough to allow for accurate recall of its functioning.

Research Sample

The research sample consisted of 94 project teams (612 individuals) working in 79 high-tech companies located in the greater New York, USA, metropolitan area. The teams were located in 8 industrial sectors including manufacturing, telecommunications, pharmaceuticals, financial services, computer hardware/software, government defense, consumer products, and consumer electronics. Each of the participating companies had enrolled one or more of their middle managers in the executive masters of technology management program at a leading technological university located in the northeastern U.S. These middle managers served as the primary company contact. Each of the company contacts identified a project team located within their company to participate in the study. Participation on the part of the project team was voluntary. The projects represented by these teams were varied and included tasks such as new product, new technology, and new services development, vendor selection for military contracts, customer service, and order fulfillment.

Data Collection Methods

Data was collected by means of two survey instruments: (1) a team information sheet, and (2) a team member survey. The information sheet was designed to collect objective, descriptive data about the team (for example, team size and project duration) and was completed by the team leader. The team member survey was designed to collect data about the team’s dynamics and interactions and was e-mailed to all team members including the team leader. Of the 97 project teams identified for study, we received 97 information sheets (one from each team leader) and team member surveys from multiple team members. For 3 of the 97 teams, less than 30% of the team responded to the team member survey, and as a result, were eliminated from the study. This yielded a sample size of 94 teams, from which we received 612 individual team member surveys. For 74 (79%) of the teams in the sample, 50% or more of the team responded to the team member survey. The remaining teams had a minimum response rate of 30% to the team member survey. The within-team response rate was 68% (6.5 average responses per team (612 responses/94 teams)/ 9.5 average team size reported by team leader). Thus, we are confident that the data collected represented the collective views of the team.

Measures

Team Size

Team size was measured by a single item on the team information sheet that asked the team leader to record the total number of team members involved in the project.

Functional Membership Diversity

In the information sheet, the leader was asked to report the total number of unique functional areas represented on the team. Thus, if multiple members belonged to the same functional area (for example, finance), the team leader would count them as one (note that in the calculation, leaders were instructed to associate each member with only one primary functional area). We then calculated functional membership diversity by dividing the number of functional areas represented on the team by team size. For example, if a team leader reported that there was 5 functional areas on the team, that would mean that the diversity measure would be equal to 1 if there were 5 members on the team (that is, each member represented a different functional area); .83 if there were 6 members on the team (5/6) (that is, two members shared the same functional area and the rest were from different functional areas); and .71 if there were 7 members on the team (5/7), and so on. This measure was adapted from Blau’s (1977) measure of diversity, which considers the number of functional areas represented by the team’s membership relative to proportion of the team. Like Blau (1977), a value of 1 represented the highest level of diversity and all values were positive (see descriptive statistics in Table 17-1).

Team Member Turnover

Team member turnover was measured on the team information sheet by asking the team leader the following question: “What percent of turnover has there been in team membership?” The team leader chose from 5 different percentage ranges: 1 = (0–20%); 2 = (21–40%); 3 = (41–60%); 4 = (61–80%); and 5 = (81–100%).

Goal Uncertainty

Two items adapted from the research by McComb, Green, and Compton (1999) were used to measure goal clarity. The items were included on the team member survey and included the reverse scores of “the team, as a whole, has clear goals and objectives” and “upper management formally approved project goal(s).” The items were rated on a 5-point Likert scale ranging from “strongly agree” to “strongly disagree.” Cronbach’s alpha for this scale was .69. A scale mean was computed for each of the team members, which was then averaged with the other team members’ scale means to provide a scale mean for the project team.

Task Interdependence

Task interdependence was measured using two items from Gladstein (1984). The two items appeared on the team member survey and included, “team members can accomplish their tasks without information from other team members (reverse scored)” and “tasks to be performed by team members can be done independently (reverse scored).” The items were rated on a 5-point Likert scale ranging from “to no extent” to “to a very great extent.” Cronbach’s alpha for this scale was .75. A scale mean was computed for each of the team members, and then averaged together with the other team members’ scale means to calculate a scale mean for the project team.

Team-Oriented Culture

Team-oriented culture was measured using a seven-item index by Compton et al. (2003). These items were included on the team member survey. Examples included (1) “My company’s mission statement refers to the use of teams”; (2) “Managers regularly discuss the importance of teams with employees”; and (3) “Teams are widely used in my company”. The items were rated on a 5-point Likert scale ranging from “strongly disagree” to “strongly agree.” Cronbach’s alpha for the cognitive conflict scale was .72. Each team member’s responses were averaged together to derive a scale mean, and all members’ scale means were averaged together to derive a scale mean for the project team.

Team-Based Rewards

Team rewards was measured using a single item adapted from the research by Van de Ven and Ferry (1980). This item was included on the team member survey and read, “When performance goals for this team are attained, how likely is it that team as a whole is rewarded or recognized for their collective achievements?” The item was rated on a 5-point Likert scale ranging from “not at all likely” to “almost a certainty.” Team members’ responses were averaged together to derive a scale mean for the project team.

Trust

Trust was measured using two items, including “overall, team members are very trustworthy” and “I could rely on those with whom I worked on the team.” All items were rated on a 5-point Likert scale, ranging from strongly disagree to strongly agree. Cronbach’s alpha for this scale was .81. Team members’ responses were aggregated and averaged to provide a scale mean for the project team.

Behavioral Integration

Behavioral integration was measured using a 5-item scale used in the research Mooney and Sonnenfeld (2001). Typical items included, “team members are mutually responsible for decision” and “team members have a clear understanding of the issues and needs of each other.” All items were rated on a 7-point Likert scale ranging from “strongly disagree” to “strongly agree.” Cronbach’s alpha for the behavioral integration scale was .85. Team members’ responses were aggregated and averaged to provide a scale mean for the project team.

Cognitive Conflict

Cognitive conflict was measured by four items used in the research by Pelled et al. (1999) and Jehn (1994). The items were included in the team member survey. Examples of items are, “How often do members of your team disagree about how things should be done?” and “To what extent are the arguments in your team task-related?” All items were rated on a 5-point Likert scale ranging from “none” to “a great deal.” Cronbach’s alpha for the cognitive conflict scale was .79. A scale mean was computed for each of the team’s members. These mean scores were then aggregated and averaged to provide a project team scale mean.

Affective Conflict

Affective conflict was also measured by four items used in the research by Pelled et al. (1999) and Jehn (1994). The items were included in the team member survey. Sample items are, “To what extent are personality clashes evident on your team?” and “How much jealousy or rivalry is there among members of your team?” All items were rated on a 5-point Likert scale, ranging from “none” to “a great deal.” Cronbach’s alpha for the affective conflict scale was .88. Team members’ responses were aggregated and averaged to provide a scale mean for the project team.

Data Aggregation

As noted above, we averaged each member’s responses to all of the items for each construct to derive a scale mean for that member. Justification for this was that the coefficient alpha for each of the measured constructs was greater than 0.7, indicating acceptable reliability (Nunnally 1978). The coefficient alpha for goal uncertainty (a = .69) was somewhat lower than desired, indicating that this measure may not be measuring a unified construct.

The unit of analysis in this study was the project team. Thus, for measures where questionnaire data provided by individual team members were averaged together, we tested the appropriateness of such aggregation by assessing whether the between-group variance was greater than the within-group variance. The F-ratio was significant (p < .001) for each of the measured variables where aggregation was used. This result indicates that the between-group variance is large relative to the within-group variance for each of the measured variables, and aggregation to the team level appears justified.