Confidence Intervals and P-Values

Abstract

Reporting the results of medical research studies with confidence intervals (CIS) is becoming increasingly recognized. In Thailand, CI has not been widely used. This is partly due to the fact that the majority of medital investigators have had little chance to refresh their biostatistics or research methodology in recent years. As a result the conventional hypothesis testing, which leads to the report of p-values is still predominantly practices. In this paper the meaning of CI is explained and its characteristics described, particularly in comparison to p-values. This paper aims at helping researchers to interpret Cls correctly when reading journal articles, and to know when to apply the CI to report their own studies appropriately. It is also directed at bridging the gap between medical investigators and bio statisticians to some extent. Once these goals are achieved, we shell expect to hear less of such a question as "Is that statistically significant?" and more of ".......... what's the confidence interval associated with that estimate?"