Logical dependencies refer to hidden interconnections among source files that are changed together in order to address an issue or change in the system. In this study we propose six metrics for logical dependency using heuristics on change time. We evaluated our proposed metrics using data from three different software companies. We also built defect prediction models with our proposed logical dependency metrics, commonly used baseline metrics, as well as combination of both. Result of our empirical studies shows that our proposed metrics are capable of capturing the characteristics of dependency among software components. It also shows the logical dependency metrics improve the performance of defect prediction models.