ABSTRACT
A hypothesis is an unproved theory that is formulated as a starting point for an investigation. In practice, we assess the probability that the effect we found would have occurred if the null hypothesis were true. If the probability is low, it follows that the effect may be due to the effectiveness of the treatment–or possibly some other cause. In order to make this assessment, we need to calculate a test statistic and use this to determine the probability (P-value). A P-value of less than our chosen threshold of significance does not prove the null hypothesis to be true–it merely demonstrates insufficient evidence to reject it. There is always an element of uncertainty when using a P-value to decide whether or not to reject null hypothesis. Although P-values are routinely calculated, there is a strength of feeling that confidence intervals may be a better way of testing hypotheses, since they show an estimate of where the true value actually lies.
