What Statistical Test is Appropriate?
The following is the outline of the factors relevant to the choice of statistical tests and a set of three charts that may be used to guide your selection of a test.
Table of Contents
Choosing the right statistical test depends on
- Nature of the data
- Sample characteristics
- Inferences to be made
A consideration of the nature of data includes
- Number of variables
- Not for the entire study, but for the specific question at hand
- Type of data
- Numerical, continuous
- Dichotomous, categorical information
- Rank-order or ordinal
A consideration of the sample characteristics includes
- Number of groups
- Sample type
- Normal distribution (parametric) or not (non-parametric)
- Independent or dependent
A consideration of the inferences to be made includes
- Data represent the population
- The group means are different
- There is a relationship between variables
Before choosing a statistical test, ask
- How many variables?
- How many groups?
- Is the distribution of data normal?
- Are the samples (groups) independent?
- What is your hypothesis or research question?
- Is the data continuous, ordinal, or categorical?
- In situations where one variable is studied, the chart below may guide your selection of statistical tests.
In situations where two variables are studied, the chart below may guide your selection of statistical tests.
In situations where three or more variables are studied, the chart below may guide your selection of statistical tests.
In summary,
- Statistical significance indicates the probability that results were chance findings
- The choice of a statistical test depends on the data, sample characteristics, and research question