Corporate Governance in Russia - Dissertation Example

A total of 74 companies was analysed, 34 companies from LSE, 4 from NYSE and 36 listed on RTS. The sample size was calculated from a Web-based sample size calculator using the following parameters : (1) a margin of error of 7%; (2) a confidence level of 95%; (3) a population size of 5,580; and (4) response distribution of 10%. The population size of 5,580 is the total number of companies listed with the Russian Trading System (297), the New York Stock Exchange (2,317) and the London Stock Exchange (2,966).

The minimum recommended sample was 70 but for contingency, this number was increased by 5%, hence the actual sample size used was 74. Companies which were listed with LSE and NYSE are categorised as class listed (CL). These are the companies are listed abroad, numbering 38. The non-class listed companies (NCL) are those companies that are listed only with RTS in Russia. The list of the companies and a screenshot of the output from the Web-based sample size calculator can be found in the Appendix.. The test is repeated until all the outliers are deleted.

Grubbs test works on the principle that with the outliers deleted, data tend to be normally distributed (Thompson and Lowthian, 2011). In this regard, use of Grubbs test requires prudence in estimating normality of the distribution in the dataset. Moreover, the test may not be applied for a small sample size of six or less since repeated iterations alter the chances of detecting outliers (Thompson and Lowthian, 2011). In the case of this research, CL and NCL data sets made the use of the Grubb’s test impossible, because it detected too many outliers, because CL firms tend to be large and well-established, also the specific environment in which firms operate would influence their board characteristics and availability of data.

Considering the big information availability difference of the treatment and benchmark populations comparison between those independent samples can be problematic. Log base 10 Further, logarithm was applied on operating revenue and number of employees. The most common description of log or logarithm of a number represents the exponent by which a fixed number, called the base, has to be exponentiated to generate the fixed number (Bland, 2007). For the current research common logarithms (logs to base 10) are useful in a several ways.

First of all, they simplify the data output for further calculations. Secondly, log transformation is applicable to data in where the residuals tend to assume bigger values as the values of the dependent variable increases. The danger in this type of scenario is that the error or change in the value of an outcome variable is a percentage and not an absolute value. Hence, similar percentage