|
Chapter 6Statistical Methods and Measurement2008-Guide to Advanced Empirical Software Engineering 3299771.3299772, BF01324126Chapter 6
Statistical Methods and Measurement
Jarrett Rosenberg
Abstract
Useful ways of measuring software engineering phenomena have to address two challenges defining realistic and valid metrics that can feasibly be collected under the constraints and time pressures of real-world software development contexts, and determining valid and accurate ways of analysing the resulting data to guide decisions. Too often, the difficulties of addressing the first challenge mean that the second is given little attention. The purpose of this chapter is to present different techniques for the definition and analysis of metrics such as product quality data. Specifically, statistical issues in the definition and application of metrics are presented with reference to software engineering examples.
1. Introduction
Measurement is ubiquitous in software engineering, whether for management, quality assurance, or research purposes. Effectively creating and using measurements is critical to success in these areas, yet there is much confusion and misunderstanding about the best way in which to define, collect, and utilize them. This chapter discusses the purpose of measurement and statistical analysis in software engineering research and development, and the problems researchers and practitioners face in using these methods effectively rather than a “how-to,” it is a
“when-to.” Section 2 discusses some fundamental issues in measurement and the context of measurement. A number of the issues in this section are discussed in the ISO/IEC 15939 standard, Information Technology – Software Measurement
Process. Section 3 discusses two basic aspects of creating effective measures metric definition and metric evaluation. Sections 4 and 5 covers methods for description, comparison, and prediction for simultaneous and successive measurements, respectively, whether categorical or numeric. Section 6 returns to the context of measurement in discussing the important topic of data quality.
155
F. Shull et al. (eds, Guide to Advanced Empirical Software Engineering.
© Springer 2008
156 J. Rosenberg Share with your friends: |
|
|
