Choosing a statistical test
DOI link for Choosing a statistical test
Choosing a statistical test book
This chapter guides the researcher on which statistical tests to use with particular kinds of data and for specific purposes, i.e. to address fitness for purpose. The chapter indicates different considerations which researchers must bear in mind when selecting the most appropriate tests, including: the number of samples, the types of data used, choosing the correct statistics, ‘safety checks’ and ensuring that the assumptions of the tests have been met. The chapter provides several tables to guide the researcher in making choices of statistic, for example, taking account of the scale of the data, the number of samples, whether the samples are independent or related, what the researcher wishes to investigate (e.g. description, correlation, difference, effect size, prediction of values or effects of one or more variables on a dependent variable, grouping of people or variables into underlying factors). The chapter presents two graphics of flow charts to guide the researcher in choosing the correct statistic, one for non-parametric data and the other for parametric data.