V. R. Basili, F. Shull, and F. Lanubile. Using experiments to build a body of knowledge. In NASA Softw. Eng. Wshop., pages 265--282, Dec. 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....size and complexity, and there is no difference in their understandibility by the subjects. Feedback Between Treatments. The subjects did not receive feedback about how well they were performing after the first treatment. This alleviates some of the problems that may be introduced due to learning [4]. If subjects do not receive feedback then they are less likely to continue applying the practices they learned during the first treatment. Power Analysis. Ideally, researchers should perform a power analysis before conducting a study to ensure that their experimental design will find a ....

....approach for defect detection in design documents. The design documents were specified in the UML. 16 For example, in previous experiments where two documents were inspected in a repeated measures design, it was noted that there was a potential for a document by treatment confounding effect [3][4][39] By ensuring document equivalence, this threat can be diminished. 22 Our results indicate that the effectiveness of teams using PBR is greater than when using CBR. Furthermore, we found that the cost per defect ratio using PBR is smaller than CBR during the defect detection phase of ....

[Article contains additional citation context not shown here]

V. Basili, F. Shull, and F. Lanubile. Using Experiments to Build a Body of Knowledge. Technical Report, University of Maryland, CS-TR-3983, 1998.

....size and complexity, and there is no difference in their understandibility by the subjects. Feedback Between Treatments. The subjects did not receive feedback about how well they were performing after the first treatment. This alleviates some of the problems that may be introduced due to learning (Basilie et al. 1998). If subjects do not receive feedback then they are less likely to continue applying the practices they learned during the first treatment. Power Analysis. Ideally, researchers should perform a power analysis before conducting a study to ensure that their experimental design will find a ....

.... or through its interaction with the treatment) 17 We presented evidence 17 For example, in previous experiments where two documents were inspected in a repeated measures design, it was noted that there was a potential for a document by treatment confounding effect (Basili et al. 1996, Basili et al. 1998; Laitenberger et al. 1999) By ensuring document equivalence, this threat can be diminished. 34 showing that the two documents were equivalent in size, and in terms object oriented structural properties that have been demonstrated to have an impact on fault proneness through increased ....

[Article contains additional citation context not shown here]

Basili, V., Shull, F., and Lanubile, F., 1998. Using Experiments to Build a Body of Knowledge. Technical Report, University of Maryland, CS-TR-3983.

....main effect. This means that the interaction cannot be estimated separately from the group effect. However, it has been argued in the past that this interaction is not important because all modules come from the same application domain, which is the application domain that the subjects work in [2][83]. Effect of Intact Groups. As noted earlier, the groups in our study were intact (i.e. they were not formed randomly) The particular situation where there is an inability to assign subjects to groups randomly in a counterbalanced design was discussed by Campbell and Stanley [8] In this design ....

....plausible that subjects do not perform the CBR and PBR reading techniques but revert to their usual technique that they use in everyday practice. This may occur, for example, if they are faced with familiar documents (i.e. documents from their own application domain within their own organisation) [83]. This is particularly an issue given that the subjects are professional software engineers who do have everyday practices. As alluded to earlier, with PBR it is possible to check this explicitly by examining the intermediate artefacts that are turned in. We did, and determined that the subjects ....

[Article contains additional citation context not shown here]

V. Basili, F. Shull, and F. Lanubile. Using Experiments to Build a Body of Knowledge. Technical Report, University of Maryland, CS-TR-3983, 1998.

....However, it has to be examined whether the analysis has to be complemented by other approaches, such as simulation. An experiment family has to allow to systematically derive empirical studies in multiple environments whose results can be aggregated later on. Previous work on experiment families [7] tried to summarize similar experiments after they were conducted; that is, the family was created after the experiments were run. This involves all the problems described in the previous section; for example, that studies are conducted and reported very differently. In contrast, our approach ....

Basili, V. R., Shull, F., Lanubile, F., 1998, Using Experiments to Build a Body of Knowledge, IEEE Transactions on Software Engineering, 25(4): 456-474.

No context found.

V. R. Basili, F. Shull, and F. Lanubile. Using experiments to build a body of knowledge. In NASA Softw. Eng. Wshop., pages 265--282, Dec. 1999.

No context found.

V. R. Basili, F. Shull, and F. Lanubile. Using experiments to build a body of knowledge. In NASA Softw. Eng. Workshop, pages 265--282, Dec. 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC