Efficiency, Incentives, and Transformational Leadership: Understanding Collaboration Preferences in the Public Sector

Empirical analysis 
Linear regression 
Descriptive statistics and zero order correlations (available in the 
Appendix
) 
were calculated for the dependent, independent, and control variables of this 
286
CAMPBELL 

---

study. Mean values of the main variables of interest range from 3.02 
(performance-based incentives) to 3.75 (willingness to collaborate). As 
hypothesized, both transformational leadership and efficiency orientation 
intensity are positively correlated with the dependent variable. Use of 
performance-based incentives, however, is not correlated with collaboration 
preferences at a statistically significant level. 
Table 1 
shows the results of a series of ordinary least squares regressions 
predicting employee willingness to engage in inter-organizational collabor-
ation. Models 1 and 2, respectively, show coefficients for the predictors with 
ministry-level effects uncontrolled and included. Both models are homosce-
dastic. The inclusion of the fixed effects raises the average variance inflation 
factor modestly from 1.20 to 1.69, and adds an additional 2% to the adjusted 
R
2 
of the model (
p 
< 0.01). As can be seen in the table, coefficients and signifi-
cance levels experience only minor changes due to the inclusion of the fixed 
effects, suggesting that unobserved ministry-level heterogeneity accounts for 
little variance in the dependent variable. Demographic controls are not related 
to the dependent variable at statistically significant levels, while inter- 
organizational task interdependence, as expected, is positively related. 
The transformational leadership, efficiency orientation intensity, and 
performance-based incentives variables were mean-centered prior to model 
inclusion and generation of the interaction terms in order to improve the 
interpretability of the coefficients (Dalal & Zickar, 2012).