The multiattribute evaluation of program-planning contractors

This note describes how the theory of measurable multiattribute value functions was used to evaluate the proposals of three competing subcontractors for a program-planning project undertaken by Los Alamos National Laboratory. The purpose of the program-planning project was to develop a methodology for the evaluation of alternative long-range strategies for the achievement of controlled thermonuclear fusion. A measurable multiattribute value function was developed based on eleven criteria. Eight judges made evaluations on each of these criteria, and their judgments were synthesized to identify the winning subcontractor. While we emphasize details associated with this specific application, we expect the problem structure and methodology to be amenable to other contractor-and proposal-evaluation efforts after only minor modifications.     

Religious homogamy and marital duration among those who file for divorce in California, 1966-1971

The hypothesis that having a common religion is associated with more stable marriages is tested using California data on divorce for the period 1966-1971. The results confirm the hypothesis, and the authors note that "religious homogamy among Jewish couples is associated with longer [marriage] duration than any other group. Couples who report no religious affiliation appear to be at greatest risk of early filing for divorce. The religious groupings include the Jewish, the Conservative Protestant, the Liberal Protestant, the Roman Catholic and those with no religious affiliation."     

Patient progress modelling and results in Wilm's Tumour

The report describes the development of a mathematical model of the progress of patients with Wilm's Tumour using the methodology developed by Jackson & Aspden [1, 2]. The model incorporates two prognostic factors which are shown to influence patient progress. These are the stage of the tumour and the age of the patient at presentation. The model proved adequate to represent an initial set of patient data, and comparison runs were performed with data from a later time period. These runs highlighted changes in patient progress between the two groups and illustrated the usefulness of the model in this type of analysis. The analysis also underlined the poor prognosis of patients over 5 years old who relapsed.