Principles of Statistics in Respiratory Research: Part 5. An Introduction to Statistical Analysis

Statistical analyses are of three general types: descriptions, tests of hypotheses, or estimations. A description of a set of measurements can be made numerically or graphically by the appropriate statistics or histogram. A test of an hypothesis has a number of steps: establishing an hypothesis, choosing and computing a test statistic, determining a rejection region for the test statistic, and making a conclusion in which the researcher must consider the possibility and the probability of an erroneous conclusion. An estimation is used when, instead of testing a set of parameters, the researcher may want to predict the value of a parameter or the difference between parameters. To do this, the researcher would compute a confidence interval consisting of a point estimate and an interval estimate. The concept of estimation can be extended to control for factors that may influence the response variable; this kind of estimation is a multiple regression analysis.