Erasmus University Rotterdam Erasmus School of Economics Master Accounting, Auditing and Control Master's Thesis Accounting, Auditing & Control Successful-Efforts
Download 0.83 Mb.
|
- Navigate this page:
- Conclusion
Testing the explanatory powersIn the previous paragraphs eight regression models are tested. The last test that has to be performed is a paired t-test. Each model has an explanatory power and that explanatory power can be linked to and explanatory power of another model. An example of a link is that the earnings model is tested for firms that used the cash-expense method for disclosing their R&D expenditures and firms that use the successful-efforts method. The main research question states that the successful efforts method is more value relevant for disclosing R&D expenditures than the cash-expense method. By paring the explanatory powers a t-test gives an answer on the main research question. The figure below gives an overview of the paired relations.
The above relations will be tested using a paired t-test. Due to the low number of observations the p-value for accepting the alternative hypothesis will be 0,1. The hypotheses are as follows: H0 µsuccessful-efforts = µcash-expense Ha µsuccessful-efforts > µcash-expense The performed t-test is a two tailed test. The test described in the hypotheses above is an one tailed test. The t-test showed a p-value of 0,098. This 0,098 is two tailed, so each tail has a p-value of 0,049. This p- value of 0,049 is lower than the critical p-value of 0,1. This means that hypothesis H0 is wrong and Ha is accepted. In other words the explanatory powers of the successful-efforts are significant higher than those from the cash-expense method. Those results are provided in the SPSS output below.
This chapter doesn’t answer the main research question. The research question will be answered in the next chapter.
This chapter described the statistical results from the regression models. This chapter begun with the seventh sub-question: “What result came from the regression models constructed in chapter five “Research design”?” First for all regression models the variables were checked on the assumptions normality, homoscedasticity, linearity and multicollinearity. Most variables showed a normal distribution of the observations. Some didn’t show a normal distribution, but the central limit theory gives base to meet the assumption of normality. The second and third assumption homoscedasticity and linearity are also met. There are some outliers but those outliers are not disturbing the research concerning the sample size. The fourth assumption is non multicollinearity. Not all models meet the assumption of non multicollinearity. Some variables showed a variance inflation factors above 10, which means a strong correlation between variables. This correlation is expected because many variables are dependent upon another. The multicollinearity is a limitation that should be kept in mind during the analyses of the next chapter, but it isn’t a large limitation. The regression models tested in this chapter provided explanatory powers between 0.007 to 0.902. The first set of models tested is the book value models. The firms that use the cash-expense method showed low explanatory powers and the variables were not significant. The firms that used the successful-efforts model showed higher explanatory power than firms that use the cash-expense method. The model with the absolute variables showed the most significant variables. The second set of models is the earnings models. The first model tested included firms that used the cash expense method. The variables have a significant relation with the returns and the explanatory power was for the scaled models 17% and for the absolute models 46%. The second model tested was for firms that used the successful-efforts method. The variables in the absolute model are significant, but the variables in the scaled model are not significant. The explanatory powers are 19% for the scaled model and 88% for the absolute model. The last test performed is a paired T-test where the explanatory powers of the successful-efforts models showed to be significant higher than the explanatory powers of the cash-expense models.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||