5.2 Prior meta-analytic research

The studies in chapter 4 derive their conclusion from a single dataset and are therefore characterised as primary research (Glass 1976: 3). These literature reviews have some drawbacks: ‘selective inclusion of studies, often based on the reviewer’s own impressionistic view of the quality of the study; differential subjective weighting of studies in the interpretation of a set of findings; misleading interpretations of study findings; failure to examine characteristics of the studies as potential explanations for disparate or consistent results across studies; and failure to examine moderating variables in the relationship under examination’ (Wolf 1986: 10). As an alternative to a literature review, a meta-analysis can be conducted that does not suffer as much from these drawbacks. ‘Meta-analysis refers to the analysis of analyses […] the statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings. It connotes a rigorous alternative to the casual, narrative discussions of research studies which typify our attempts to make sense of the rapidly expanding research literature’ (Glass 1976: 3).

The relationship between board independence and performance has been the subject of a number of meta-analytic studies (Dalton and Dalton 2011: 406-407): Dalton et al. (1998), Rhoades et al. (2000) and Wagner et al. (1998). Dalton et al. (1998) found 51 studies with 159 samples and an aggregated sample size of 40,160. All samples together show a modest significant positive correlation between board independence and diverse measures of performance (1998: 279). Certain moderator effects, such as company size, performance measure or type of board composition measure, show that these results are not robust. Wagner et al. (1998) conducted a comparable meta-analysis of 29 studies with 63 samples and an aggregated sample size of 13,619 company years. They reported a significant positive correlation between board independence and financial performance (1998: 663-666). When they divided the total sample into subgroups, based on different sorts of performance measures, most subgroups still reported a significant correlation between board independence and performance. The differences in correlations between the different subgroups of performance measures are not significant, which means that they fail to produce evidence that certain relationships are stronger than others (i.e. one performance measure is not more correlated with board independence than another performance measure). The same applies to a division between different forms of board composition. Although the different groups all report significant positive relationships, the differences between these board composition measures are not significant.

A third meta-analysis was conducted by Rhoades et al. (2000). They found 30 studies with 37 samples and an aggregated sample size of 7,644. In contrast to the other two studies, Rhoades et al. did not find any significant correlations between independence and performance (2000: 81-85). The split into different subsamples did not result in significant correlations either. The fact that Rhoades et al. did not find any significant relationships, whereas Dalton et al. and especially Wagner et al. did find these significant correlations, is notable. The techniques in all three articles are quite similar and there is overlap between the articles used as well. Rhoades et al. have 22 and 19 studies (out of 30) in common with Dalton et al. and Wagner et al., respectively. Wagner et al. have 26 studies (out of 29) in common with Dalton et al. Besides the overlap in studies used, all three meta-analyses have some ‘unique’ studies that might have influenced the difference in results. However, the fact that ninety per cent (26 out of 29) of the studies of Wagner et al. are also used by Dalton et al. should raise some eyebrows about why they have ignored the other studies of Dalton et al.

These three meta-analyses do not reach consensus about the (significant) relationship between the two variables of interest: board independence and financial performance. Although initiatives on corporate governance are numerous, which becomes apparent in part II of this study, the studies used in the three meta-analytical studies are rather outdated; the most recent study investigated dates from 1996. In addition, the studies primarily conducted research on American data and put less emphasis on research from outside the United States. Therefore, this chapter applies a meta-analysis to the relationship between board independence and financial performance, using data from the United States as well as Europe and other western countries. It focuses on research that was published between January 2000 and March 2011 in order to focus on publications that use data from after the start of corporate governance reforms at the beginning of this century. Only correlation relationships are used and not results from regression analyses. Although information on regression relationships is gathered for this study, these results have not been included in a final analysis due to the many complications that arise when regression slopes are cumulated (Becker and Wu 2007).