Statistical Test Explorer
Every test the statistical test guidance knows, all at once. Answer the questions in any order and the tests that no longer fit fade out. The number beside each answer is how many tests would remain if you chose it.
Type of study
Number of measurements*
Independent variable
Number of variables*
Scale*
Number of levels
Dependent variable
Number of variables*
Scale*
Number of levels
What do these questions mean?
Type of study
This tool distinguishes between 5 major types of designs that may apply to a research question.
Describe data for groups : choose this design when the RQ only asks for a summary of the data (e.g., mean or median).Compare group with hypothetical value or distribution : choose this design when the RQ asks about how the data collected for a group of subjects compares to a value generated by different means (e.g., previous research, a theory) or if it asks if the data follows a certain distribution (e.g., normal distribution).Compare two or more groups : choose this design when the RQ asks how the data collected for identifiable, separate groups of subjects compare between these groups (e.g., in experimental design between control and treatment groups). Use for causal inferences.Study repeated measures designs : choose when the RQ asks how the same group of participants compares across measurement occasions for one or more measures (e.g., pre-post or time series designs). Can be used for causal inferences.Study associations between variables : choose when the RQ asks if a pattern of association that may exist between two or more variables is due to chance or not (e.g., one variable grows with the growth of another variable), but no information about causality is sought. Number of measurements
The number of measurements question asks about how many times the same group of subjects have been measured. The options are:
One : only once (e.g., cross-sectional studies)Two : exactly two times (e.g., pre-post studies)Two or more : at least two times but can be more (e.g., time series) Number of variables
How many independent (IV) or dependent (DV) variables are being included in the RQ. To note that there may be RQs that do not have independent variables.
None : available only for the independent variables; choose in case the RQ does not have an IV.One : choose if the RQ has one IV only;Two : choose if the RQ includes exactly two IVs./Two or more : choose if the RQ includes at least two IVs. Scale
Choose how the RQ variable(s) (either IV or DV) are measured. The four scales provide increasingly more information about the data (e.g., the ordinal scale offers more information than the nominal scale). If the RQ includes more than one variable, chose select the scale offering the least amount of information.
In order:
Nominal : classifies unordered, mutually exclusive categories (e.g., gender as male or female). Offers the least amount of information about the data.Ordinal : ranking hierarchy of levels (e.g., letter grades). Provides more information that a nominal variable by offering a ranking (order) of the levels, following a specific, known rule.Interval : improves on the ordinal scale by adding equal differences between the levels, with real life equal meaning or counterpart, with one downside, it has no true zero point (e.g., the Celsius scale has a value of 0, but that value does not mean there is no heat present).Ratio : similar to interval scales, but with a true zero point (e.g., the system of inches), which gives these scales the ability to reason about ratios between scale values. Number of Levels
The number of levels for the variable(s), IV or DV. If the RQ has more than one variable, choose the option matching the variable with the highest number of levels.
N/A : choose when the number of levels are not relevant (e.g., if the RQ does not have an IV). Two : chose when the variable has exactly two levels (e.g., a variable that has only yes/no answers).Two or more : choose when the variable has two or more levels (e.g., letter grades).Many : choose when a variable has many levels (e.g., a scale between 0 and 100 has 100 levels).
2 of 46 tests fitStill needed for suggestions: Number of measurements, IV number of variables, IV scale, DV number of variables, DV scale
Kendall Correlation
Close
Non-parametric test used to measure the strength and direction of the association between two ordinal, interval, or ratio variables, based on how many pairs of observations are ranked in the same order. An alternative to the Spearman Correlation Coefficient for small samples or many tied ranks. Assumes a monotonic relationship between the variables.
Use when: As for Spearman's coefficient; often preferred with small samples or many tied ranks.
Levels: Favored for variables with more than two levels. When one of the variables has only two, the point-biserial correlation is the better choice.
Alternative to: Pearson Correlation Coefficient . Both apply; start with it. “Use when” says when this test fits instead, and what it compares.
Test card, explanations, and examples → (opens in a new tab)
Faded: It is used to study association between variables, not to study repeated measures designs.
Number of measurements
One
IV number of variables
One
IV scale
Ordinal, Interval, Ratio
IV number of levels
Two or more, Many
DV number of variables
One
DV scale
Ordinal, Interval, Ratio
DV number of levels
Two or more, Many