This essay analyzes the way to improve the linkage between a theoretical model and empirical analysis in political science. Based on Goertz (2005), first, I exemplify that classifying concepts into three levels and rigorously conceptualizing the relationship among basic and secondary-level concepts contribute to assessing the validity of her theoretical model. Second, I show that conceptual analysis also facilitates a case selection. According to the Possibility Principle, only cases where the outcome of interest is possible should be treated as negative ones; cases there the outcome is impossible should be considered uninformative and irrelevant. Probing set relations among explanatory and explained factors allow a researcher to distinguish negative cases from irrelevant cases. Third, through critically appraising my own research, I demonstrate that fuzzy-set analysis can heighten concept-measurement consistency. When inventing a theoretical model through natural language, a researcher tends to develop a complex model with categorical set relations. However, empirical data usually have continuous values in reality. Fuzzy-set analysis adapts continuous data to set relations and enables a researcher to analyze necessary/sufficient conditions among them.