Using firm-specific regressions, I show that earnings response coefficient differ across firms. However, there is no evidence of differential earnings response coefficient to a certain earnings announcement time. By switching to a different announcement time from its preferred time, a firm does not gain a softer market reaction. I compare research results from a firm-specific method and from a pooled time-series and cross-sectional method and demonstrate that they differ significantly due to large heterogeneity across firms. I suggest that researchers should adopt a firm-specific approach to avoid misleading results and to achieve improved estimations.