However, because this is a single bank study, the generalizability of results should be tested in the future. The study should be replicated in other contexts using other indicators of performance, like profitability and customer satisfaction. A variety of ways exist in which the calibration sample can be drawn; for example, outstanding service quality performers can be selected and their commitment or other variables computed to identify ideal profiles. Also, the calibration sample can be taken at any level, which could improve the coefficient of determination, R2. Some researchers prefer the top 5% while others prefer the top 20%. These can be adjusted to fit with the purpose of the exercise. Future research can also empirically compare profile deviation methods with existing statistical techniques.