IC assets as mediating variables – hypothesis 1

In order to find out, if hypothesis one: KM practices have a positive impact on organizational performance, when being mediated by IC assets, was supported, organizational performance was regressed on the variables. Nine sub-hypotheses were tested.

Hypothesis 1a) Leadership -> IC assets -> Organizational performance tested

Leadership had a strong, direct impact (β = 0.41) on organizational performance (Table 6.). The impact of leadership on the mediators is shown in Appendix 5. These regression analyses showed that all parameter estimates had a high value. When IC assets were regressed on leadership, they got values of: human capital (β = 0.49), structural capital (β = 0.50), and relational capital (β = 0.39). The impact of leadership on IC assets mediators was strong and significant (P = .0001).

When these IC assets mediators were one after another added to the regression analysis, the impact of leadership on value weakened (Figure 10.). Instead of showing an effect similar to the direct effect (β = 0.41), the impact of leadership on value, got values of β = 0.32 (human capital as a mediator), β = 0.36 (structural capital as a mediator), and β = 0.32 (relational capital as a mediator). The model, and the impact of leadership on value, were still significant (P = .0001).

Analysis 1 Analysis 2

Analysis 3 Figure 10. Leadership mediated by IC assets

The values of the parameter estimates were compared by calculating: leadership without a mediator (direct impact) – leadership with a mediator = indirect effect, and the results were:

1) Leadership -> human capital -> value: 0.41 – 0.32 = 0.09 2) Leadership -> structural capital -> value: 0.41 – 0.36 = 0.05 3) Leadership -> relational capital -> value: 0.41 – 0.32 = 0.09

The results show the unexplained part of the direct effect when a mediator is added to the model. So to say, this much of the direct impact on organizational performance was lost due to an IC assets mediator, and could be explained by mediation.

What needs to be pointed out is that the impact of human capital and structural capital were not significant in the model, and these results on mediation need to be critically observed. If they are excluded from the mediation point of view, the only IC asset that possibly mediated leadership was relational capital (0.09). The coefficients of determination of the models were good however: 1) 27,9 %, 2) 26,2

% and 3) 32,2 % (adjusted), and the t-tests measuring the goodness of fit were on an adequate level: human capital 2,95; structural capital 2,17 and relational capital 4,80.

These findings implicated that leadership did have a positive impact on value, whether being mediated by IC assets or not. Full mediation was far from being confimed, but a direct effect was found. This was shown as a change in the size of the coefficients. Because the significance of the effect of leadership on value did not weaken, when IC assets functioned as mediators, mediation was not supported from the significance point of view in any of these sub-hypotheses.

Hypothesis 1b) Training and development -> IC assets -> Organizational performance tested

Training and development showed a rather strong, direct impact (β = 0.24) on organizational performance (Table 6.). The size of the coefficient was smaller than that of leadership. The impact of training and development on the mediators is shown in Appendix 5. All of the parameter estimates had a rather high value. When IC assets were regressed on training and development, they got values of: human capital (β = 0.29), structural capital (β = 0.34), and relational capital (β = 0.29). The impact of training and development on IC assets mediators was strong and significant (P = .0001).

When these three IC assets mediators were added to the regression analysis one by one, the impact of training and development on value got weaker than the direct impact (Figure 11.). When the direct impact of training and development on value was β = 0.24, the mediated impact was β = 0.14 (human capital), β = 0.18 (structural capital), and β = 0.14 (relational capital). The impact of development on value was less significant when human capital (P = .0006) or relational capital (P = .0005) functioned as the mediator.

Analysis 4 Analysis 5

Analysis 6 Figure 11. Development mediated by IC assets

When the values of the parameter estimates were compared by calculating:

development without a mediator (direct impact) – development with a mediator = indirect effect, the results were:

4) Development -> human capital -> value: 0.24 – 0.14 = 0.10 5) Development -> structural capital -> value: 0.24 – 0.18 = 0.06 6) Development -> relational capital -> value: 0.24 – 0.14 = 0.10

According to this analysis 0.06 or 0.10 of the positive impact of training and development on value was explaned by an IC assets mediator. The impact of structural capital on value was a little bit less significant than the other mediators impacts. The coefficients of determination of the models were: 4) 21,4 %, 5) 17,1 % and 6) 24,1 % (adjusted), which were much lower than those of the previous analyses, where leadership was the predictor variable. The t-tests measuring the goodness of fit were higher than those in 1a) and on an adequate level: human capital 5,30; structural capital 3,65 and relational capital 5,98.

These findings implicated that training and development did have a positive impact on value, whether being mediated by IC assets or not. Full mediation was far from being confimed, but instead, partial mediation was found when both; the size of the coefficients and the significance of the variables, were analyzed. Sub-hypothesis four and six were supported. The significance of the effect of development on organizational performance, got somewhat weaker when human capital (4) or relational capital (6) functioned as a mediator. From the significance point of view, the less significant the formerly significant effect is, more assuring mediation there is. Because the partial mediation pointed at the same direction, this mediation effect was also complementary.

Hypothesis 1c) Compensation -> IC assets -> Organizational performance tested

Compensation had a positive, direct impact (β = 0.16) on organizational performance (Table 6.). This impact was much weaker than those of the two previous predictor variables. The impact of compensation on the mediators is shown

in Appendix 5. When IC assets were regressed on compensation, they got values of: β = 0.14 (human capital), β = 0.13 (structural capital), and β = 0.10 (relational capital). The significance of the relationships altered.

When these three IC assets mediators were added to the regression analysis one by one, the impact of compensation on value was weaker (Figure 12.) than the direct impact without a mediator. The impact of compensation on value mediated by human capital was (β = 0.11), structural capital (β = 0.14), and relational capital (β

= 0.12). The impact of compensation on value got more significant when IC assets mediated it, except for human capital that resulted in a somewhat lower significance.

These confusing results might be a sign of multicollinearity that was described in Chapter 4.3.

Analysis 7 Analysis 8

Analysis 9 Figure 12. Compensation mediated by IC assets

When the values of the parameter estimates were compared by calculating:

compensation without a mediator (direct impact) – compensation with a mediator = indirect effect, the results were:

7) Compensation -> human capital -> value: 0.16 – 0.11 = 0.05 8) Compensation -> structural capital -> value: 0.16 – 0.14 = 0.02 9) Compensation -> relational capital -> value: 0.16 – 0.12 = 0.04

According to this analysis 0.02 to 0.05 of the effect of compensation on value was explained by an IC assets mediator. There was positive mediation supported by the analyses, but it was very weak. Compensation did not have a significant impact on structural capital nor relational capital, which indicated that there might be other factors impacting the effect. The coefficients of determination of the models were however good: 4) 22,5 %, 5) 18,6 % and 6) 25,4 % (adjusted), and a bit higher than those of 1b), but lower than those of 1a). The t-tests measuring the goodness of fit were surprisingly high, exceeding the results of both 1a) and 1b). The values of the t-tests were: human capital 6,30; structural capital 4,87 and relational capital 7,03.

These findings implicated that compensation did have a positive impact on value, whether being mediated by IC assets or not. As in the previous analysis, full mediation was far from being confimed, but signs of partial mediation were found instead. Sub-hypothesis seven got supported when both; the size of the coefficients and the significance of the variables, were analyzed. The significance of the impact of human capital on organizational performance was a bit weaker due to human capital as a mediator. This supported partial mediation. Because the significance of the effect of compensation on value did not weaken when structural capital or relational capital functioned as a mediator, mediation was not supported in these cases.

To summarize hypothesis one

KM practices have a positive impact on organizational performance, when being mediated by IC assets, got supported (Appendix 7). None of the models were able to show full mediation, but sub-hypotheses 4, 6 and 7 showed partial mediation. The highest changes in the size of the coefficient (0.10) were shown in sub-hypotheses four and six. The size of the coefficient in sub-hypothesis seven changed 0.05. The biggest change in the significance level of the predictor variable then was found in sub-hypothesis four (.0006). The significance of the predictor in sub-hypothesis six was .0005 and sub-hypothesis seven .0003. These findings make mediation practically non-existent from a significance point of view, because the significance of the effects did not change substantially. When only the coefficients were compared, there were direct and indirect effects found in everyone of the nine sub-hypothesis.