i-nth logo

Authors

Jimmie L. Joseph

Abstract

The computer science literature has long espoused the virtue of using groups of programmers to locate errors in computer programs, and details the methodologies to employ in group debugging efforts. Literature also indicates that this methodology is cost effective for the firms that utilize it.

However, while much spreadsheet error detection research cites computer science as a reference discipline, no published spreadsheet error detection studies to date have examined the effects of different group sizes on the detection of spreadsheet errors. This study presents an overview of spreadsheet research and error detection research/paradigms. It then presents hypotheses concerning group effects and interactions and experimentally tests those hypotheses.

Four hypotheses and three exploratory hypotheses are tested. It was found that groups did not find more errors than individuals. Groups did find a higher net number of errors (seeded errors found minus false reports). Groups that previewed the spreadsheet as individuals found more errors and more net errors than groups that did not preview the spreadsheet.

In examining the exploratory hypotheses, it was found that an increase in group size resulted in an increase in the number of lost errors (errors detected by a member of the group during individual examination of the spreadsheet, but not reported by the group to which they were assigned in the second phase of the experiment). Also, as group size increased, the number of errors that had not been reported by any member of the group, but which were reported after the group examination of the spreadsheet, increased.

Finally, it was found that groups that had previewed the spreadsheet as individuals spent less time reviewing the spreadsheet in their group, than groups that had not previewed the spreadsheet.

Sample

Summary of results
Summary of results

This table shows a summary of the hypothesis results. All but one of the hypotheses was supported by the experimental results.

Publication

2002, Ph.D thesis, University of Pittsburgh

Full article

Not available