Logistic likelihood analysis of mediation outcomes

Dispute persists in all building and construction projects. Alternative dispute resolution methods are now commonly used as a means to resolve construction disputes. Mediation, interalia, is the popular choice in Hong Kong due to its cost-saving, flexible, speedy, confidential, and voluntary attributes. In mediation, tactics used by a mediator is central in driving desired outcomes. This paper reports a study that employs logistic regression (LR) to predict mediation outcomes respective to the tactics used. To achieve this, three main stages of work are involved. First, taxonomies of mediator tactics and mediation outcomes were developed. With these, the second stage included the development of logistic regression models each with a mediation outcome taxonomy as dependent variables and the taxonomies of tactics as independent variables. In the third stage of the study, the LR models were validated using an independent set of testing data. The LR models suggested that "win-win settlement," "progress," "improvement," and "time advantage" are responsive to mediator tactics of "ice-breaking," "trust building," "encourage for self-improve," and "process control," respectively. In addition, it is observed that these relationships are positively correlated.