Following a Non-Significant Finding (2 of 3)

---

The experimenter's belief that sleep deprivation increases reaction time should be strengthened. Nonetheless, the data are not strong enough to convince a skeptic, so no attempt should be made to publish the results. Instead, the experimenter repeated the experiment. Once again, the sleep-deprived group had slower reaction times. Based on the results of the first experiment, the experimenter conducted a one-tailed test in the second experiment. However, it was not significant, p = 0.08, one tailed.

The naive interpretation of the two experiments is that the experimenter tried twice to find a significant result and failed both times. With each failure, the strength of the experimenter's case that sleep deprivation increases reaction time is weakened.

The correct interpretation is that in two out of two experiments the sleep-deprived subjects had the slower reaction times. The experimenter's case is strengthened by each experiment. Moreover, there are methods for combining the probability values across experiments. For these two experiments, the combined probability is 0.047.