5.4.6 Realisation of the study

The corporate governance statements to be analysed were collected from the website www.company.info and the websites of the companies themselves. To ensure that the analysis was performed in an objective way a total of 14 coders analysed the corporate governance statements1 (with a minimum of 2 coders per corporate governance statement). The systematic approach of this study was guaranteed by the coders' training, the pilot study they performed, the coding form they worked with, and the explicit coding instructions. In the event of different answers among the coders a third party decided which of the given answers was considered to be correct. By means of content analysis the following main questions (concisely put) were addressed:

The reliability of these results (the intercoder reliability) was measured by means of Cohen's kappa: a standard intercoder reliability coefficient for content analysis (Neuendorf 2002) (Krippendorff 2004) that measures the level of agreement between the coders corrected for the expected agreements by change.3 The Cohen's kappa measured in this study has a value between 0.419 and 0.807 and can be considered more than sufficient4 (see table 5.4.6 below).

The aim of the present study is to add to previous research, but its limitations are also acknowledged. The level of compliance of the companies under review is measured by the contents of their corporate governance statement. Unfortunately, companies can mention in their corporate governance statement that they comply with the provisions of the applicable code but may deviate from it in practice. As stated before, material compliance with the code is hard to measure and this problem is inherent to the chosen corporate governance model. However, it is not simply the hard compliance figures that are of interest in this study, but rather the trends in the level of compliance and (the quality) of the specific explanations given by the companies. The comply or explain principle and compliance with corporate governance codes have been studied before, for example by the Dutch Corporate Governance Code Monitoring Committee (Report Dutch Monitoring Committee 2008) and for the UK by Arcot and Bruno (Arcot and Bruno 2007). However, this is the first study that, using one method, measures the application of the principle (non-compliance and reasons given for non-compliance) for three consecutive years for five EU countries simultaneously. Only the RiskMetrics Group have studied code compliance internationally, however for fewer companies per country and for one year.5 As Durisin and Puzone state: "There is an empirical gap in cross-national studies in the literature" (Durisin and Puzone 2009, p. 279). The underlying research tries to assist in closing this gap.