Does class size matter? An in-depth assessment of the effect of class size in software defect prediction

In the past 20 years, defect prediction studies have generally acknowledged effect of class size on software performance. To quantify relationship between object-oriented (OO) metrics and defects, modelling has to take into account direct, potentially indirect, effects defects. However, some shown that cannot be simply controlled or ignored, when building models. As such, there remains a question whether, when, control for size. This study provides new in-depth examination impact OO defects defect-proneness. We assess number defect-proneness in systems by employing regression-based mediation (with bootstrapping) moderation analysis investigate direct indirect count binary prediction. Our results show is not always significant all metrics. Of seven we investigated, consistently only Coupling Between Objects (CBO) defects/defect-proneness, potential Fan-out defects/defect-proneness. Other mixed results, they are but others. Based our make three recommendations. One, encourage researchers practitioners examine specific data hand through use proposed statistical mediation/moderation procedures. Two, empirical possible additional variables their models relevant. Three, procedures adopted this could used other engineering research influence mediators/moderators.