Assessing the Relationship between Prerelease Software Testing and the Number of Product Defects Discovered.
详细信息
- 作者：Nakkhongkham ; Saiyoot.
- 学历：Doctor
- 年：2011
- 导师：Ferreira,Gail,eadvisor
- 毕业院校：Northcentral University
- ISBN：9781124592237
- CBH：3452479
- Country：USA
- 语种：English
- FileSize：6143944
- Pages：151
文摘
The software testing process can be unpredictable due to deficiencies in the measurement process,resulting in poor quality of software releases. In this research study,software test effectiveness was assessed by measuring the relationship between code coverage and the number of prerelease and postrelease product defects. This study employed a quantitative,single-case study method to observe software testing activities performed on 86 software modules at a company located in the State of Hawaii. In this study,the NCover code coverage tool was used to measure the relationship between code coverage and the number of prerelease and post release defect by assessing actual lines of code written for a software module. The total prerelease and postrelease product defects were counted. Nonparametric statistical tests called Spearman rank correlation coefficients and Wilcoxon signed-rank tests were used to analyze the data. Results showed a) a significant positive association exists rho84) ＝ .38,p <； .001) between code coverage and the number of prerelease defects,b) a significant negative association exists rho84) ＝ - .51,p <； .001) between code coverage and the number of postrelease defects,and c) a significant difference exists z ＝ -7.88,p <； .001) between the number of prerelease and postrelease defects. The associations found in this quantitative study between code coverage and the number of defects,together with the difference in number of defects between prerelease-postrelease defects matched-pairs,are important software test-quality indicators capable for use in describing the state of software test effectiveness as well as the quality of the product release. Furthermore,the critical point of 85% code coverage for sufficient software testing was found to be a representative number that could be used by other companies who were testing a software product of similar size and scope. Recommendations for future research include assessing other types of software modules using the same coverage testing theory methodology as a baseline for measurement techniques,as well as a quality indicator of software releases.