The Effectiveness of Using Mediation in Selected Civil Law Disputes: A Meta-Analysis

4. Summation

Much of the preceding analysis is quite dense and it is a bit difficult to see the ‘forest’ of what works in mediation for the ‘trees’ of numerous variables that are considered in various mediation studies.  This section will highlight the most relevant findings from the preceding presentation.  Choosing what to highlight is contingent upon the robustness of the statistical summaries undertaken in the meta-analysis.   

For 10 of the 17 outcome measures we were able to make a confident statement about the positive impact of a mediation program versus a comparison group reported in one of the studies we examined. Therefore, this meta-analysis does indicate some broad improvement in many outcome measures. When there is a mediation program the following outcomes are improved:

Our sample size was fairly small because not many studies exist that met our study criteria. Therefore, when a more detailed analysis of overlapping variables are undertaken the numbers with which calculations are made becomes very small. Thus, it was quite difficult to report with statistical confidence about every outcome measure since, for some outcome measures, the findings of different studies varied too widely. However, when we cannot make a statistically confident statement about a variable, it may not be variability or small sample that is the issue. It is also possible that the impact of mediation on that outcome measure is actually very low, negligible or even possibly marginally negative. This was the case in seven instances:

These ambiguous or negative results may be caused by any number of factors. All would require further evaluative research. A number of these outcomes are directly related to judicial and legal procedure (number of hearings, pre-trial conferences, number of motions). Attempting to compare them across very different jurisdictions and areas of law may have resulted in a degree of variability that does not allow for a useful comparison. While the meta-analysis showed a positive impact of mediation on actual cost savings and perceptions of costs savings, perceptions of reasonable cost savings was not. This may be because expectations of cost savings as a result of using the mediation program were higher than the actual cost savings that were realized.

It is also interesting to note that the surveys in some studies (k=6) asked people in the comparison group if they would have preferred to use mediation over the process they did use. Of the people surveyed, slightly less than half (48%) thought they would have preferred to use the mediation alternative. This ambivalent result may be due to the fact that people accommodate themselves to the situation in which they find themselves or that knowledge of what was (or would) have occurred in the mediation program with their case could only be imperfectly known (if at all).

In most cases the analysis of moderator variables was not fruitful. The majority of moderator variables which could be confidently noted as improving an outcome measure could not be compared with any statistical reliability to any other moderator variable within the same outcome measure. While these moderator variables may be most strongly associated with effective mediation programs, they may also be artefacts of sample size. Thus, not many conclusions can be drawn from them. There were limited instances where more than one moderator variable within an outcome measure could be compared. However, in the majority of these instances either the confidence interval ranges of the moderator variables overlapped (indicating a similar impact) or one of the moderator variables was a catch-all category where we could not be sure what was included. In only three instances, moderator variables could be more clearly contrasted with one another. The voluntary selection of mediation was associated with reduced staff hours to a greater degree than mandatory mediation. Case lengths are lower and measured cost savings are higher in mediation programs that are compared against a comparison group that only allows mediation ineligible cases versus those that allow mediation eligible cases. However, very small samples make drawing any firm conclusions from these comparisons difficult.

Only one of the calculations in the meta-analysis found a negative outcome of mediation. In this one instance, the meta-analysis found that that programs where there was freedom to select a mediator resulted in a lower settlement rate than the comparison group, but that mediation programs where there was no freedom to select the mediator demonstrated an improved settlement rate.