Power estimation is most useful when conducted a priori--when modifications in the intended design and/or analysis strategies can still be made. If not used a priori, however, it still can be helpful as a post hoc secondary analysis, both for consumers of research literature and for researchers themselves. In cases when the null hypothesis is not rejected at the predetermined alpha level, it may be very informative to conduct a power analysis. The researcher might discover that there was very little power to reject the null hypothesis if it were indeed false. Knowing that this is a reasonable possible explanation for the "no significant difference" conclusion could preclude abandonment of the project or premature publication of discouraging results. It also could help in the design of future, similar studies.