Abstract

How one may use descriptive statistics to give a succinct description of research data is first discussed. The probability basis of inferential statistics, namely, the random sampling distribution of the test statistic, is then introduced. The said sampling distribution is used to introduced the null-hypothesis significance-testing procedure (NHSTP). The emphasis on 'procedure' serves to highlight the fact that significance tests are about data, not about the substantive hypothesis. The distinction is made between (a) the statistical alternative hypothesis (H1) and the substantive hypothesis, (b) using NHSTP to test whether or not chance is responsible for the data and using the embeddding conditional propositions to corroborate the theory. NHSTP serves to supply the minor premise for the theory-corroboration procedure. Some conceptual difficulties with effect-size, statistical power and meta-analysis are also discusse.