ABSTRACT
Many statistics operate with dependent and independent variables (e.g. experiments using t-tests and analysis of variance, regression and multiple regression); others do not (e.g. correlational statistics, factor analysis). If one uses tests which require independent and dependent variables, great caution has to be exercised in assuming which is or is not the dependent or independent variable, and whether causality is as simple as the test assumes. Further, many statistical tests are based on linear relationships (e.g. correlation, regression and multiple regression, factor analysis) when in fact the relationships may not be linear (some software programs, e.g. SPSS, have the capability for handling nonlinear relationships). The researcher has to make a fundamental decision about whether, in fact, the relationships are linear or non-linear, and select the appropriate statistical tests with these considerations in mind. To draw these points together, the researcher will need to consider:
What scales of data are there?OO Are the data parametric or non-parametric?OO Are descriptive or inferential statistics required?OO Do dependent and independent variables need to be OO identified? Are the relationships considered to be linear or non-OO linear?
