Empirical P value is the P-value calculated for the actual observed data instead of theoretical data. In statistics, significance testing is carried out by taking up two hypotheses to explain the statistical model for the process. First is null hypothesis which gives the simplest model. Another is alternative hypothesis, which is true if null hypothesis is proved wrong. These two hypotheses are tested against each other in terms of P- value.
P –value, considering null hypothesis to be true, gives the probability of getting the test statistic at least as extreme as one that was actually observed. It tests the dataset to find out if it is unlikely to get the observed value if the null hypothesis is true. A high P-value favors null hypothesis. Normally, significance level, i.e., Type I error is used as the cutoff point. Hence if the significance level if 0.05 and if the P-value obtained is smaller than this, the null hypothesis is rejected. On the other hand, higher the P-value, higher the confidence of null hypothesis being true.
Empirical P-value uses the actual observed data to calculate the P-value and then to reject to accept null hypothesis.