In the past, much emphasis was put on the so-called significance testing. The investigator assumed a null hypothesis stating no association between the exposure and the disease (usually the real hypothesis would be the opposite of the null hypothesis). Then he/she would calculate a P value. The P value would indicate the probability of getting the data he/she found or data that were even further away from the null hypothesis (the no-effect value), given the null hypothesis was true (and other conditions). If this P value was below a given level (often < 0.05) it was said that the finding was statistically significant and the null hypothesis was rejected as a likely explanation of the data.