Statistical Analyses and Results

514 explained, commitment has one factor 67.35 of the variance explained, product and market policy has one factor 80.64 of the variance explained, financial performance has one factor 53.12 of the variance explained, and non financial performance has one factor 86.21 of the variance explained. The results confirm that the questionnaires used in this study can be categorized into their intended constructs. [Insert Table 3 here] Table 4 reveals the Pearson correlations for all variables used in this study. The two performance variables used in this study is highly correlated r = 0.792, p 0.001 indicating that they measure the same construct. [Insert Table 4 here] Table 5 presents the descriptive statistics about the variables used in this study for defender, prospector, analyzer and total sample. The mean responses for financial performance measures are: 3.96, 3.93, 3.68, and 3.79 for defender, prospector, analyzer and total sample respectively. The mean responses for non financial measures are: 3.59, 3.55, 2.90, and 3.47 for defender, prospector, analyzer and total sample respectively. In terms of performance evaluation, the mean responses are: 3.30, 5.43, 4.88, and 4.06 for defender, prospector, analyzer, and total sample respectively. The mean responses for compensation are: 3.67, 5.20, 4.47, and 4.16 for defender, prospector, analyzer, and total sample respectively. The mean responses for compensation are: 3.67, 5.20, 4.47, and 4.16 for defender, prospector, analyzer, and total sample respectively. The mean responses for communication are: 4.96, 5.28, 5.01, and 5.05 for defender, prospector, analyzer, and total sample respectively. The mean responses for conflict resolution are: 3.39, 5.12, 4.02, and 3.91 for defender, prospector, analyzer, and total sample respectively. The mean responses for commitment are: 4.97, 3.30, 3.69, and 3.76 for defender, prospector, analyzer, and 515 total sample respectively. The mean responses for product and market policy are: 4.31, 4.34, 4.44, and 4.35 for defender, prospector, analyzer, and total sample respectively. Table 5 also presents the descriptive statistics of the misfit constructs using both financial and non-financial performance measures for the crucial control systems MISFIT_C and non-crucial control systems MISFIT_NC. The means of MISFIT_C for financial measures are: 1.79, 4.27, 2.22, and 2.19 for defender, prospector, analyzer and total sample respectively. The means of MISFIT_NC for financial measures are: 0.01, 0.03, 0.21, and 0.03 for defender, prospector, analyzer and total sample respectively. The means of MISFIT_C for non financial measures are: 1.13, 4.52, 2.26, and 2.04 for defender, prospector, analyzer and total sample respectively. The means of MISFIT_NC for non financial measures are: 0.01, 0.05, 0.27, and 0.04 for defender, prospector, analyzer and total sample respectively. [Insert Table 5 here] Table 6 present the results of the OLS regression analyses for financial performance measure as the dependent variable for each type of strategy 69 . The results indicate that performance evaluation β = 0.440, p 0.01, compensation β = 0.191, p 0.05, conflict resolution β = 0.β4β, p 0.05, and commitment β = 0.176, p 0.01, are significantly related to performance for defenders. For prospectors, compensation β = -0.β46, p 0.10, commitment β = 0.58γ, p 0.01, and product and market policy β = 0.488, p 0.01, are significantly related to performance. For analyzers, only commitment β = 0.777, p 0.05 is significantly related to performance. [Insert Table 6 here] Table 7 present the results of the OLS regression analyses for non financial performance measure as the dependent variable for each type of strategy 70 . The results indicate that performance evaluation β = 0.γ05, p 0.05, compensation β = 69 For the sake of completeness, we also include the OLS regression for the total sample. 70 For the sake of completeness, we also include the OLS regression for the total sample. 516 0.ββ6, p 0.10, conflict resolution β = 0.β41, p 0.10, and commitment β = 0.β0β, p 0.05, are significantly related to performance for defenders. For prospectors and analyzers, only commitment β = 0.750, p 0.01; and β = 0.688, p 0.10, respectively is significantly related to performance. [Insert Table 7 here] Table 8 summarizes the results of the correlation analyses between MISFIT_C and performance and also between MISFIT_NC and performance for financial performance. It also reports the results of the test for the difference in the magnitude of the correlation coefficients 71 . Hypothesis H1a predicts that a misfit between business strategy and the critical control systems will have a negative and significant impact on financial performance. The results indicate that the misfits of strategy and the crucial control variables MISFIT_C have negative and significant correlations with financial performance for all types of strategy r = -0.916, p 0.01 for defenders; r = -0.958, p 0.01 for prospectors; and r = -0.689, p 0.01 for analyzer. The results are consistent with hypothesis H1a. Hypothesis H2a posits that a misfit between business strategy and the non critical control systems will not have a significant effect on financial performance. The results reveal that the correlations between MISFIT_NC and financial performance are not significant for all types of strategy. The results confirm hypothesis H2a. Hypothesis H3a expects that the correlation between MISFIT_C and financial performance will be significantly more negative than the correlation between 71 We use the procedure proposed by Chen and Popovich 2002 to test the difference in magnitude of the correlation coefficients. The following formula is used to perform this test: 3 3 1      ns s rns rs n n z z z where z rs and z rs are the z –values of the correlation coefficients for the MISFIT_C and MISFIT_NC, respectively obtained from the following formula: 1 1 log 5 . r r X z e r    . 517 MISFIT_NC and financial performance. The results show that the correlations between MISFIT_C and financial performance are significantly more negative than the correlation between MISFIT_NC and financial performance for all types of strategies z = 8.495, p 0.01 for defenders; z = 5.321, p 0.01 for prospectors; and z = 1.984, p 0.05 for analyzers. The results support hypothesis H3a. [Insert Table 8 here] Table 9 reveals the results of the correlation analyses between MISFIT_C and non financial performance and also between MISFIT_NC and non financial performance. It also reports the results of the test for the difference in the magnitude of the correlation coefficients 72 . Hypothesis H1b predicts that a misfit between business strategy and the critical control systems will have a negative and significant impact on non financial performance. The results indicate that the misfits of strategy and the crucial control variables MISFIT_C have negative and significant correlations with non financial performance for all types of strategy r = -0.826, p 0.01 for defenders; r = -0.711, p 0.01 for prospectors; and r = -0.528, p 0.01 for analyzer. The results are consistent with hypothesis H1a. Hypothesis H2b posits that a misfit between business strategy and the non critical control systems will not have a significant effect on non financial performance. The results reveal that the correlations between MISFIT_NC and non financial performance are not significant, except for prospectors where the correlation is negative and significant r = -0.209, p 0.05. The results provide some support to hypothesis H2a. Hypothesis H3b expects that the correlation between MISFIT_C and non financial performance will be significantly more negative than the correlation between 72 We use the same procedure to test the difference in magnitude of the correlation coefficients as discussed earlier for financial performance. 518 MISFIT_NC and non financial performance. The results show that the correlations between MISFIT_C and non financial performance are significantly more negative than the correlation between MISFIT_NC and non financial performance for defenders and prospectors z = 5.927, p 0.01 and z = 2.194, p 0.05 respectively. For analyzers, however, the correlation difference is not statistically significant. The results provide some support to hypothesis H3b. [Insert Table 9 here]

5. Discussions, Limitations, and Implications for Future Research

The notion of strategy-control system misfits is a central theme in management accounting research utilizing a contingency approach and the performance implications of strategy-control system misfits are intuitively appealing. However, little research has been conducted to support this proposition. In particular, no studies have considered using relative weights of different control systems. This is important since the notion of equal weight is generally considered untenable. In this study, we derive the weights by performing OLS regressions and use the standardized beta weights of the regression equation of control system variables on performance for each type of strategy. The results indicate that the strategy-control system misfit for crucial control system variables has a significant negative effect on performance. By contrast, the strategy-control system misfits for non-crucial control system variables do not affect performance negatively except for the marginally significant effect on performance for prospectors when non-financial measures of performance are used as the dependent variable. More importantly, the magnitude of the correlations between misfits and financial performance for the crucial control variables are significantly more negative than the correlations between misfit and performance for the non-crucial control variables for all types of strategies. Furthermore, the correlations between misfit and non-financial performance for crucial control variables are more negative than those for non crucial control variables. The magnitude 519 of the differences of the two correlations, however, is only statistically significant for the defender strategy. It is interesting to note that not all of the correlations between misfits and performance for the non-crucial control systems are close to zero. As shown in Table 9, the correlation between misfit and non-financial performance for the non-crucial control variables is negative and marginally significant for prospectors. This unexpected result has two implications. First, deviation from the ideal profile in terms of non-crucial control system could have a negative and significant impact on performance. Second, there is the need to compare between the magnitude of correlations between misfits and performance for the critical and non-critical control systems to provide evidence that the misfits in terms of crucial control variables are more damaging than the misfits in terms of non-crucial control variables. As such, our study might help firms improving their performance by focusing the deployment of their limited resources on control systems that are crucial for the companies to thrive and succeed, given their chosen strategy. The results of this study should be interpreted in light of three limitations. First, the misfit construct was derived empirically by comparing the control systems used by high performing firms and those of the sample firms. It might be that the high performing firms use control systems that are not consistent with the theoretical prescriptions of the strategy- control system fit. However, the theoretically derived ―ideal profile‖ is still debatable and the operational task of specifying such profile with numerical scores along a set of MCS is a difficult task Venkatraman Prescott, 1990. We leave this for future research. Second, we use data from banking industry which is known for its highly regulated and tight government control. Future studies might use data from different industry to enhance the generalizability of our results.