336 Empirical Research in Software Engineering
of results in those situations that are different from the subjects and settings ofthe study.
Forexample, a study which evaluates the effectiveness of machine learning methods to iden-
tify fault-prone classes using data mined from open source software would have external
validity concerns regarding whether the results computed using machine learning algo-
rithms on a given data set hold true for industrial software data sets or for other open-source
data sets. A study with high external validity is favorable as its conclusions can be broadly
applied in different scenarios, which are valid across the study domain (Wright et al. 2010).
Thus, the results obtained from a study with more number of ...