An Experiment Family to Investigate the Defect Detection Effect of Tool-Support for Requirements Inspection

Halling, Michael; Biffl, Stefan; GrÃ¼nbacher, Paul

doi:10.1109/METRIC.2003.1232474

An Experiment Family to Investigate the Defect Detection Effect of Tool-Support for Requirements Inspection

by Michael Halling, Stefan Biffl, Paul GrÃ¼nbacher

Abstract:

The inspection of software products can help to find defects early in the development process and to gather valuable information on product quality. An inspection is rather resource-intensive and involves several tedious tasks like navigating, sorting, or checking. Tool support is thus hoped to increase effectiveness and efficiency. However, little empirical work is available that directly compares paper-based (i.e., manual) and tool-based software inspections. Existing reports on tool support for inspection generally tend to focus on code inspections while little can be found on requirements or design inspection. This paper reports on an experiment family: two experiments on paper-based inspection and a third experiment to empirically investigate the effect of tool support regarding defect detection effectiveness and inspection effort in an academic environment with 40 subjects. Main results of the experiment family are: (a) The effectiveness is similar for manual and tool-supported inspections; (b) the inspection effort and defect overlap decreased significantly with tool support, while (c) efficiency increased considerably with tool support.

Reference:

Michael Halling, Stefan Biffl, Paul GrÃ¼nbacher, "An Experiment Family to Investigate the Defect Detection Effect of Tool-Support for Requirements Inspection", IEEE Computer Society, pp. 278-285, 2003.

Bibtex Entry:

@Conference{Halling2003a,
  Title                    = {An Experiment Family to Investigate the Defect Detection Effect of Tool-Support for Requirements Inspection},
  Author                   = {Michael Halling and Stefan Biffl and Paul GrÃ¼nbacher},
  Booktitle                = {Proceedings 9th IEEE Int'l Software Metrics Symposium (METRICS 2003), 3-5 September, Sydney, Australia},
  Year                     = {2003},
  Pages                    = {278-285},
  Publisher                = {IEEE Computer Society},

  Abstract                 = {The inspection of software products can help to find defects early in the development process and to gather valuable information on product quality. An inspection is rather resource-intensive and involves several tedious tasks like navigating, sorting, or checking. Tool support is thus hoped to increase effectiveness and efficiency. However, little empirical work is available that directly compares paper-based (i.e., manual) and tool-based software inspections. Existing reports on tool support for inspection generally tend to focus on code inspections while little can be found on requirements or design inspection. This paper reports on an experiment family: two experiments on paper-based inspection and a third experiment to empirically investigate the effect of tool support regarding defect detection effectiveness and inspection effort in an academic environment with 40 subjects. Main results of the experiment family are: (a) The effectiveness is similar for manual and tool-supported inspections; (b) the inspection effort and defect overlap decreased significantly with tool support, while (c) efficiency increased considerably with tool support.},
  Doi                      = {10.1109/METRIC.2003.1232474},
  ISBN                     = {0-7695-1987-3},
  Researchr                = {http://researchr.org/publication/HallingBG03}
}