Comments on "Researcher bias: The use of machine learning in software defect prediction"
Shepperd et al. (2014) find that the reported performance of a defect prediction model shares a strong relationship with the group of researchers who construct the models. In this paper, we perform an alternative investigation of Shepperd et al. (2014)’s data. We observe that (a) researcher group shares a strong association with the dataset and metric families that are used to build a model; (b) the strong association among the explanatory variables introduces a large amount of interference when interpreting the impact of the researcher group on model performance; and (c) after mitigating the interference, we find that the researcher group has a smaller impact than the metric family. These observations lead us to conclude that the relationship between the researcher group and the performance of a defect prediction model may have more to do with the tendency of researchers to reuse experimental components (e.g., datasets and metrics). We recommend that researchers experiment with a broader selection of datasets and metrics to combat potential bias in their results.