BRANDO BReakpoint Analysis with Nonparametric Data Option


  1. Emond, E.J.
  2. Turnbull, A.E.
Corporate Authors
Defence R&D Canada - Centre for Operational Research and Analysis, Ottawa ON (CAN);Operational Research Div, Ottawa ONT (CAN) Central Operational Research Team
This paper reports on the mathematical details and software solution to the Operational Research problem of comparing the central values of multiple datasets using nonparametric statistics. The work is an extension of the iterative post-hoc analysis method described by Emond and Massel (2003). The original methodology required an assumption of Normality for all of the datasets, an assumption that is often not met. An additional analysis method has now been developed in which the data are ranked allowing for a nonparametric solution. Using a proxy for the likelihood function on the ranked datasets, this method finds the most likely separation points between them after a nonparametric analysis of variance has indicated that differences exist. In addition to the theoretical methodology and explanation, examples are given to demonstrate the practicality of the process. Of particular interest is an example which illustrates the application of the methodology to the analysis of ordered categorical data often found in surveys.

Il y a un résumé en français ici.

nonparametric data;BRANDO;SPAM;breakpoint analysis;post-hoc;ANOVA;separation point analysis;statistically significantly different
Report Number
DRDC-CORA-TM-2006-40 — Technical Memorandum
Date of publication
01 Nov 2006
Number of Pages
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: