• Keine Ergebnisse gefunden

Chapter 3 - Active Performance in Research and Development:

3.3. Results

As project management style is a team level variable, we examined intraclass correlation (ICC 1) and agreement for multiple item scales (rwg.j) to justify aggregation of responses of individual team members to the project team level (Bliese, 2006). These analyses were conducted with the subset of 41 project teams on which more than one team member had participated. ICC 1 and rwg.j for the three subdimensions of project management style are listed in Table 3.1. For the overall measure of project management style ICC 1 was .23 indicating that 23% of the variance in responses was due to differences between teams.

The mean rwg.j value was .88.

Goal-orientations. Goal orientations were assessed with the work domain self-report instrument by VandeWalle (1997). Each dimension was measured with five items on a five-point Likert scale. For mastery orientation, Cronbach’s alpha was .83. An example item is: “I enjoy challenging and difficult tasks at work where I’ll learn new skills”. For performance-prove orientation Cronbach’s alpha was .68; an example item is: “I’m concerned with showing that I can perform better than others”. For performance-avoid goal orientation an example item is: “I prefer to avoid situations at work where I might perform poorly”.

Cronbach’s alpha for the scale was .85

Active performance. Supervisors assessed team member’s active performance on five items of the personal initiative scale developed and validated by Frese, Fay, Hilburger, and Leng (1997). An example item is: “This employee actively attacks problems”. Cronbach’s alpha for the five-point Likert scale was .90. Items of the scale measure to what extent employees in general perform in an active rather than passive way. Frese (2008) proposed active performance as a concept that integrates personal initiative and related constructs such as engagement and taking charge. We therefore use the term active performance rather than personal initiative.

Table 3.2

Means (M), standard deviations (SD) and intercorrelations of all variables

Variable M SD 1 2 3 4 5 6 7 8

1 Age 31.16 7.80 -

2 Gender 1.60 0.48 .08 -

3 Tenure 4.20 6.50 .77** .01 -

4 Mastery orientation 3.70 0.57 .03 -.01 .01 - 5 Performance-prove

orientation 2.96 0.63 -.10 -.01 -.01 .05 - 6 Performance-avoid

orientation 2.23 0.76 .02 .03 .01 -.13 .44** -

7 Exploration

orientation 3.71 0.54 -.01 .20 -.02 .29** -.14 -.21* -

8 Active performance 3.78 0.76 -.14 -.03 -.15 .25** .13 -.06 .14 - 9 Project management

stylea 2.98 0.48 .15 .03 .22 .25 .02 .02 .18 .10

-Note: Sample size is N = 111.

a Project management style is a team level variable. Individual level variables were aggregated to the team level to calculate correlations with project management style. High values indicate a planned project management style; low values indicate an emergent project management style. Sample size for these correlations is N = 49.

* p ≤ .05. ** p ≤ .01. (two-sided test of significance).

Table 3.3 presents three hierarchical linear models to test hypotheses 1 to 5. After entering individual level predictors in Model 1 and 2, we entered the team level predictor project management style in Model 3. In Model 1 exploration orientation was regressed on goal orientations after controlling for age, gender and tenure. Performance-prove goal orientation was not included, because it was unrelated to the dependent variables of interest and was not of theoretical interest in the present study. In Model 2 we predicted active performance by individual level variables. In model 3 team level predictors were added.

Table 3.3

Hierarchical linear models Independent

variables Model 1 Model 2 Model 3

Dependent variables

Exploration orientation

Active performance

Active performance Individual level

Age -0.00 (.01) -0.01 (.01) -0.01 (.01)

Gender 0.23 (.10) -0.01 (.01) -0.04 (.15)

Tenure 0.00 (.01) -0.02 (.02) -0.02 (.02)

Mastery orientation 0.26 (.09)** 0.33 (.13)* 0.30 (.13)*

Performance-avoid

orientation -0.12 (.06) -0.03 (.09) -0.01 (.09)

Exploration orientation a 0.08 (.14) -0.04 (.14)

Team level

Planned vs. emergent project management style

(main effect)b 0.02 (.17)

Planned vs. emergent project management style

(cross-level interaction) b -0.76 (.31)*

Model R2 .10 .13 .19

Note. The values are unstandardized parameter estimates for the regression weights (γ).

Standard errors are indicated in parenthesis. Sample sizes are NI = 111 at the individual level and NT = 49 at the team level. p ≤ .10 * p ≤ .05 ** p ≤ .01 (two-sided).

a High values indicate an orientation towards exploration

b High values indicate a planned project management style; low values indicate an emergent project management style.

In support of Hypothesis 1, mastery orientation explained significant variance in exploration orientation (Model 1: γ = 0.26, p < .01). In Model 2 mastery orientation

incrementally predicted active performance, providing support for Hypothesis 2 (γ = 0.33, p <

.01). Thus, employees high in mastery-orientation were oriented towards exploration and showed higher degrees of active performance. The relationship between mastery orientation and active performance remained significant if team level predictors were included (Model 3). Concerning the hypothesized negative relationship between performance-avoid

orientation and exploration orientation (Hypothesis 3), model 2 shows that performance-avoid orientation was negatively related to exploration orientation (γ = -0.12, p = .054). If we

included the 67 participants for whom supervisor ratings were not available in this analysis, the negative relationship between performance-avoid orientation and exploration orientation was significant at the p < .05 level. As can be seen in model 2, no support was found for Hypothesis 4. Performance-avoid orientation was not significantly related to active performance.

Model 3 in Table 3.3 provides the test of hypothesis 5 that postulated a moderating effect of project management style on the relationship between exploration orientation and active performance. In this model, the team-level variable project management style was added as a predictor of active performance (main effect) and as a predictor of the relationship between exploration orientation and active performance (cross-level moderation). In support of hypothesis 5, the coefficient for the cross-level moderation effect of project management style was significant (γ = -0.76, p = .02).

Figure 3.2 illustrates the cross-level moderation. The relationship between exploration orientation and active performance was positive if project management style was emergent and negative if project management style was planned. Thus, people high in exploration orientation were high in active performance if project management style was emergent and low in active performance if project management style was planned. The reverse was true for people with a low exploration orientation, i.e. people with focus on exploitation. Region-of-significant tests (Preacher et al., 2006) for different values of the moderator showed that the slope for exploration orientation and active performance was significantly positive for values

< 0.4 standard deviations below the mean of the moderator (i.e. a more emergent project management style). Slopes were significantly negative for values that were > 1.42 standard deviations above the mean of the moderator (i.e. a more planned project management style).

We conducted additional analyses using the subscales of project management style rather that the composite variable. Crosslevel moderation was significant for managerial control (γ = -0.66, p < 01) and project directiveness (γ = -0.44, p = .02). For structure, the coefficient had the same algebraic sign but was not significant (γ = -0.29, p = .14).

Figure 3.2. Cross-level moderation of project management style

In order to interpret interactions as cross-level moderation, the interaction needs to remain significant after all between-team variance is removed from the individual level predictor (Hofmann & Gavin, 1998). We therefore conducted cross-level moderation analyses after group-mean centering exploration orientation, which removes between-team variance. As group-mean centering is only possible for teams for which more than one employee had participated, the sample size was reduced to 104 individuals working in 41 project team. The cross-level moderation for this model was also significant (γ = 0.89, p = .03). Thus, team members with a higher exploration orientation as compared to the other members of their team were higher in active performance if project management style was emergent. The reverse was true if project management style was planned.