A productivity benchmarking case study using Bayesian credible intervals

Authors:
John Moses;Malcolm Farrow;Norman Parrington;Peter Smith
Affiliations:
School of Computing and Technology, University of Sunderland, Sunderland, UK SR6 0DD;Department of Mathematics and Statistics, University of Newcastle-Upon-Tyne, Newcastle-Upon-Tyne, UK NE1 7RU;School of Computing and Technology, University of Sunderland, Sunderland, UK SR6 0DD;School of Computing and Technology, University of Sunderland, Sunderland, UK SR6 0DD
Venue:
Software Quality Control
Year:
2006

Citing 3
Cited 0

Empirical studies of assumptions that underlie software cost-estimation models

Information and Software Technology
Assessing Variation in Development Effort Consistency Using a Data Source with Missing Data

Software Quality Control
Practical Statistics for Medical Research

Practical Statistics for Medical Research

Quantified Score

Hi-index	0.00

Visualization

Abstract

A productivity benchmarking case study is presented. Empirically valid evidence exists to suggest that certain project factors, such as development type and language type, influence project effort and productivity and a comparison is made taking into account these and other factors. The case study identifies a reasonably comparable set of data that was taken from a large benchmarking data repository by using the factors. This data set was then compared with the small data set presented by a company for benchmarking. The study illustrates how productivity rates might be misleading unless these factors are taken into account. Further, rather than simply giving a ratio for the company's productivity performance against the benchmark, the study shows how confidence about the company's performance can be expressed in terms of Bayesian confidence (credible) intervals for the ratio of the arithmetic means of the two data sets.