Modelling the Likelihood of Software Process Improvement: An Exploratory Study

  • Authors:

  • Khaled El-Emam;Dennis Goldenson;James McCurley;James Herbsleb

  • Affiliations:

  • National Research Council, Montreal Road, Bldg. M-50, Ottawa, Ontario, Canada K1A OR6/ khaled.el-emam@iit.nrc.ca;Software Engineering Institute, Carnegie Mellon University, Pittsburgh, PA 15213-3890, USA/ dg@sei.cmu.edu;Software Engineering Institute, Carnegie Mellon University, Pittsburgh, PA 15213-3890, USA/ jmccurle@sei.cmu.edu;Software Production Research Department, Lucent Technologies, Naperville, IL, USA/ herbsleb@research.bell-labs.com

  • Venue:
  • Empirical Software Engineering
  • Year:
  • 2001

Quantified Score

Hi-index 0.00

Visualization

Abstract

Software process assessments have become big business worldwide. They can be a powerful tool for initiating and sustaining software process improvement (SPI). However, SPI programs sometimes fail. Moreover there still are very few systematic empirical investigations about the conditions under which SPI initiatives vary in their outcomes. In this paper we present the results of a study of factors that influence the success of SPI. The data come from a sample survey of organizations that have performed assessments based on the capability maturity model for software, and was conducted from 1 to 3 years after the assessments (sufficient time had passed for changes to have taken place). The results consist of a multivariate model of the conditions (e.g., how the improvement efforts are organized and funded) that can explain the successes and failures of SPI efforts. The model is constructed using a classification tree algorithm. It identifies the most important factors that affect the outcome of SPI efforts, and describes how those factors interact with each other to influence success or failure.