Grand Canyon University Statistics Study Design Discussion

Selecting the most appropriate statistical tool for analyzing public health data depends first on they type of study conducted (Corty, 2016, p. 829). If a descriptive study was conducted, descriptive statistical tests must be used, including mean, median, and mode for interval/ratio-, ordinal-, and nominal-level data, respectively, and range, variance, and standard deviation for interval/ratio-level data only (Corty, 2016, p. 832). Other types of studies, including experimental, quasi-experimental, and correlational, use inferential statistics, which uncovers a difference or relationship between variables, and thus requires difference or relationship tests to analyze the data (Corty, 2016, p. 829). Selecting the correct difference test depends first on the number of groups or samples involved, as well as the number of independent (explanatory) variables, the sample type, and the level of measurement of the dependent variable (Corty, 2016, p. 840). All relationship tests require a single group of cases with each case being measured on two variables; selecting the correct one depends only on the level of measurement of each variable (Corty, 2016, p. 846). For example, when both variables are interval/ratio, the Pearson r is indicated, but when both are nominal, a chi-square test is most appropriate (Corty, 2016, p. 846).