Some statistical terms

This list of terms is to start you off. A much fuller glossary of terms is given in Sprinthall (2003). Degrees of freedom (d.f) gives the number of scores that are free to vary once certain restrictions have been placed on the data. To illustrate the point, when placing eggs into a standard egg box that takes six eggs ( n ), only one depression would be left for the sixth egg, allowing no ‘freedom’ about where it can be placed. Hence the degrees of freedom are ( n − 1) = 6 − 1 = 5. The larger is the sample size, the greater is the number of degrees of freedom. In statistical analysis a degree of freedom is lost for every additional treatment that is added. Effect size refers to the magnitude of an effect (or, in medical parlance, the ‘clinical significance’). Effect size and statistical significance are quite separate matters and the level of statistical significance does not, as is often supposed, directly measure the magnitude or scientific importance of the observed result. A correlation or a difference can be very small in size yet highly statistically significant, provided the sample size is large enough (see Chapter 11 for further discussion). Error: Measurement error is the combined error that results from inevitable imperfections and variability in the process of measurement.