The International Court of Justice (ICJ) consists of fifteen permanent judges, each of whom shall come from a different State. A State party can appoint an ad hoc judge if no judge of its nationality sits among the permanent judges. In accordance with the ICJ Statute, judges, either permanent or ad hoc, shall exercise their powers impartially. But in reality, can this expectation be achieved? One of the most systematic empirical researches to address this topic was done by Posner and de Figueiredo who studied the ICJ cases from 1946 to 2004. Inspired by their research, in our recent paper, “An Empirical Study of the Voting Pattern of Judges of the International Court of Justice (2005-2016)”, we have conducted a continued study on the ICJ cases between 2005 and 2016. The raw data of this study can be accessed at ResearchGate.
Our null hypothesis is that the voting pattern of the ICJ judges is unbiased. Assuming that the null hypothesis is true, decisions made by the ICJ judges are influenced only by relevant legal considerations regardless of whether the applicant or respondent is the judge’s home State or appointing State (if yes, “party State”; and otherwise, “nonparty State”) and whether the applicant or respondent to a particular case shares the same or similar political, economic or cultural levels with the judge’s home State or appointing State. We use the variables democracy, wealth and language to respectively illustrate or measure the political, economic or cultural levels of relevant States. As to methodology, we use two methods: the difference tool and the regression tool.
On the basis of the data collected, we found that nonparty judges vote for applicants or respondents about 50% of the time. In contrast, party judges vote for applicants that appoint them about 88% of the time, and vote for respondents that appoint them about 89% of the time. Hence, the probability of a judge voting in favour of the applicant/respondent increases significantly when the applicant/respondent State appoints the judge. As a result, there is substantial evidence that party judges of the ICJ vote in favour of their home States or appointing States. Furthermore, by running several regressions, we found strong evidence that ICJ judges favour applicants that appoint them and that judges favour respondents that speak the same majority languages as their States, and weaker evidence that judges favour respondents that appoint them, and more weakly or no evidence that judges favour applicants that speak the same majority languages as their States or that judges are influenced by democracy and wealth alignments. These results suggest that political and cultural factors do play a role in influencing the votes of the ICJ judges.
Notably, our results are not exactly the same as those of Posner and de Figueiredo’s, especially with regard to the influence exerted by the democracy and wealth alignments. These two variables were significant in cases between 1946 and 2004, but no longer in cases between 2005 and 2016. The diluted influence exerted by the democracy and wealth alignments on the judges’ voting pattern may not be an isolated or surprising event, considering that the world has experienced significant changes around the time of the Millennium, including the dissolution of the Soviet Union, the end of the cold war, the emergence of a multi-polar world economy, globalisation and so on. By contrast, the variable language alignment remains significant in the current research. This may in part confirm Samuel P. Huntington’s prediction in his classic work, The Clash of Civilizations and the Remaking of World Order, that “the great divisions among humankind and the dominating source of conflict will be cultural”. Nevertheless, to avoid doubt, this research has not indicated that judges are consciously biased. Neither does it suggest that the biased voting pattern of the ICJ judges will necessarily lead to unfair interpretation or application of law. All it has shown is that their voting pattern is not in line with the manners described by the null hypothesis.