Psychology Statistical Test Advisor

Choosing the right statistical test is one of the more frustrating parts of psychology research. You can have a sensible hypothesis, a carefully designed study and a perfectly respectable dataset, only to discover that deciding whether you need a t-test, ANOVA, correlation or chi-square test is an entirely separate problem.

Our free Psychology Statistical Test Advisor helps psychology and social science students narrow down the most appropriate statistical tests for their research.

Answer a few questions about your research aim, variables and study design to receive a suggested test, a possible alternative and guidance on the assumptions you should check before proceeding.

The advisor is free to use, with no account required.

How to Choose the Right Statistical Test

Statistical test selection generally begins with four questions:

1. What is your research aim? Are you comparing groups, investigating a relationship, testing an association or predicting an outcome?

2. What type of data do you have? Are your variables continuous, ordinal or categorical?

3. How is your study designed? Are your observations independent, paired or repeated?

4. What assumptions are reasonable? Does your data meet the requirements of the statistical procedure you're considering?

The answers help narrow down the possibilities, although more complex research designs may require additional considerations.

Common Statistical Tests in Psychology

t-Tests: Comparing Two Means

A t-test is commonly used when comparing means.

An independent-samples t-test compares two separate groups, such as memory performance between participants who slept normally and participants who experienced sleep restriction.

A paired-samples t-test compares related observations, such as anxiety scores measured before and after an intervention in the same participants.

The distinction between independent and related observations is essential.

ANOVA: Comparing Several Groups or Conditions

Analysis of variance (ANOVA) is commonly used to compare means across three or more groups or conditions.

A one-way ANOVA examines the effects of a single factor. Factorial ANOVA allows researchers to investigate more than one factor and their interactions.

Repeated-measures ANOVA is designed for related measurements, although more complex repeated-measures designs may benefit from mixed-effects models.

A statistically significant omnibus ANOVA result does not automatically tell you which specific groups differ. Appropriate follow-up comparisons may be needed.

Correlation: Investigating Relationships

Correlation examines the direction and strength of association between variables.

Pearson's correlation is generally used for linear associations between continuous variables.

Spearman's rank correlation is useful for ordinal data or relationships that can be described as monotonic.

Kendall's tau provides another rank-based approach and may be useful in particular situations involving tied ranks.

Correlation does not establish causation. Two variables can be associated without one directly causing the other.

Chi-Square: Analysing Categorical Data

Chi-square tests are commonly used with categorical frequencies.

A chi-square test of independence examines whether two categorical variables are associated.

A chi-square goodness-of-fit test compares observed category frequencies with a specified expected distribution.

For paired categorical measurements, alternative methods such as McNemar's test may be more suitable. The correct choice depends on whether the responses are binary, nominal, independent or repeated.

Regression: Predicting Outcomes

Regression models examine how one or more predictors relate to an outcome.

Linear regression is commonly used for continuous outcomes, while logistic regression is used for binary outcomes.

Other models, including multinomial and ordinal logistic regression, accommodate different types of categorical outcomes.

Choosing an appropriate model also requires consideration of sample size, model assumptions, influential observations and the number of predictors.

Parametric vs Non-Parametric Tests

Parametric and non-parametric methods differ in their statistical assumptions and the questions they answer.

Common alternatives include:

These are not universally interchangeable procedures.

For example, the Mann–Whitney U test should not automatically be interpreted as a comparison of medians, and a failed normality test does not necessarily mean an analysis must become non-parametric.

The appropriate alternative depends on which assumption is problematic and what the researcher intends to test.

How to Use the Statistical Test Advisor

Select your main research goal from the dropdown menu.

The tool will display additional questions relevant to your chosen pathway, including outcome type, number of groups, dependence between observations and statistical assumptions.

Select Get suggested test to receive a shortlist.

The recommendation includes a possible statistical approach, an alternative worth investigating and a checklist of methodological considerations.

You can copy the summary into your research planning notes or discuss it with your supervisor.

Can the Advisor Tell Me Exactly Which Test to Use?

Not definitively.

The free advisor is intended as a decision-support tool. It cannot inspect your raw data, verify statistical assumptions or fully assess every research design.

Factors such as outliers, missing observations, unequal variances, nested data, interactions and sampling methods may change which analysis is appropriate.

Use the recommendation as a starting point, then check your statistical output, relevant methodological guidance and any requirements set by your course or supervisor.

Need More Detailed Statistical Guidance?

Simply Put Premium includes the advanced Psychology Statistical Test Advisor alongside further tools for research design, statistical assumptions, data interpretation and academic writing.

Premium also includes the Simply Put Core tools for APA tables, graphs, results reporting and reference organisation, plus additional study resources and an ad-free member experience.

For students who already know which statistical test they need and primarily want help presenting their findings, the Simply Put Core Pack provides the four main academic formatting tools as a one-off purchase.

The free Statistical Test Advisor remains available for quick test shortlisting and research planning, without an account or rewarded-ad unlock.