neutrality scale ask about judges’ ability to resist political pressure, judicial independence, national
bias, and the role played by political factors in judicial decisions. The alpha reliability coefficient
for this scale is .64. Finally, we include demographic characteristics: education, gender, and age.
The results of the logistic regression are summarized in Table 8 (Model 1). Perceived
fairness of the decisions is one of the strongest predictors of the overall perceived fairness of the
ICTY. The odds that the respondents who evaluated the ICTY decisions as unfair would regard
the ICTY in general as unfair as well are estimated to be 12.5 times as high as those of the
respondents who evaluated the ICTY decisions as fair, all other covariates being equal.
the views of the fairness of the procedures directly affected the respondents’ views of the ICTY’s
overall fairness. However, the effect of the procedural fairness is smaller than the effect of the
decision fairness: the odds that the respondents who evaluated the ICTY procedures as unfair
would regard the ICTY in general as unfair as well are estimated to be 3.9 times as high as those
of the respondents who evaluated the ICTY procedures as fair, all other covariates being equal.
Finally, the respondents who selected retribution were more likely to view the ICTY as unfair
than the respondents who selected other purposes of punishment (Table 8). None of the remaining
variables in Model 1 are statistically significant in their influence at the conventional .05 level.
Once we entered the timing of the survey (pre-2000 and post-2000) into the model
(Model 2, Table 8), it turned out to be a significant factor. In fact, the odds that the post-2000
respondents would regard the ICTY in general as unfair are estimated to be 2.4 times as high as
those of the pre-2000 respondents. With the timing of the survey in the model, the effects of
fairness of the decisions, fairness of the procedures, and punishment purpose remained strong. In
addition, compared to the oldest respondents (over 51 years of age), the youngest respondents
(18-35 years of age) were less likely to evaluate the ICTY as unfair.
In Model 3 we added the political scale in the model (Table 8). The views about how
politically independent the ICTY is are positively related to the respondents’ views of the ICTY’s
fairness in general. In particular, a one unit increase on the political scale (on which higher scores
indicate lesser degree of political independence) increases the odds 2.0 times that the respondents
would evaluate the ICTY as unfair. The effect of the views about the fairness of the decision
remains, but the importance of the respondents’ perceptions of the fairness of the procedures
disappears. It seems that viewing the ICTY as lacking political independence carries substantially
more weight in the respondents’ overall views of the ICTY’s fairness than the views about the
1
The odds effects are obtained by exponentiation of the logistic coefficients.
Convention