Internal validity concerns with the rigour of the study design. In our study, the threats are related to the data extraction process, the selection of the factors that influence code review, and the validity of the results. While we provided details on the data extraction, data filtering and any heuristics used in the study, we also validated our findings with the WebKit developers and reviewers. We contacted individuals from Google, Apple, BlackBerry, and Intel and received insights into their internal processes (as discussed in V-A).
Our empirical study is the subject to external validity; we can not generalize our findings to say that both organizational
and personal factors affect code review in all open source projects. While we compared WebKit’s code review process
with the one of Mozilla Firefox and found that its patch lifecycle is similar to open source projects, the fact that
WebKit is being developed by competing organizations makes it an interesting case yet a rather obvious exception. Hence,
more studies on similar projects are needed.
Statistical conclusion validity refers to the ability to make an accurate assessment of whether independent and dependent
variables are related and about the strength of that relationship. In order to determine whether relationships between variables are statistically significant or not we performed null hypothesis testing. We also applied appropriate statistical tests (analysis of variance, post-hoc testing, and Spearman’s correlation).