Univariate analysis examines one variable, bivariate analysis examines two together, and multivariate analysis examines several in one analysis. The right choice depends on the question you want to answer and the kinds of data you have—not on which method sounds most advanced. One terminology wrinkle: some fields use “multivariate” broadly, while others reserve it for analyses with multiple outcome variables.
What do univariate, bivariate, and multivariate mean?
The labels describe how many variables an analysis considers together. They also suggest the kind of question it can answer: a distribution, a relationship or group difference, or a model that considers several variables at once.
As an Amazon Associate I earn from qualifying purchases.
| Type | Variables considered together | Typical question | What the result describes |
|---|---|---|---|
| Univariate | One | What does this variable look like? | The variable’s distribution, such as category counts or numerical center and spread. |
| Bivariate | Two | How are these two variables related, or do groups differ? | A pairwise association, comparison, or difference. |
| Multivariate or multivariable | Several | How do several variables relate to an outcome, or how do multiple outcomes behave together? | A joint or adjusted relationship, depending on the model and terminology used. |
These are broad categories, not a complete list of statistical methods. The same dataset can support all three kinds of analysis, depending on the question.
What does univariate analysis do?
Univariate analysis looks at one variable on its own. It is useful for understanding what values occur, how common they are, and how spread out they are. It does not, by itself, show how that variable relates to another.
#1 Best Overall
- For a categorical variable, use counts or proportions—for example, how many students took an in-person, hybrid, or online course.
- For a numerical variable, summarize appropriate measures of center and spread and use a suitable display—for example, examining the distribution of exam scores.
Describing a variable first can reveal its basic pattern and help you decide what questions to ask next. It is a useful starting point, but not a requirement for every project.
What does bivariate analysis do?
Bivariate analysis considers two variables together. Depending on the question and data, it may describe an association, compare groups, or test whether evidence supports a difference or relationship. The variables may be numerical, categorical, or a combination.
Rank #2
- This guide is a perfect overview for the topics covered in introductory statistics courses.
- Two numerical variables: Explore how study hours and exam scores vary together using an appropriate plot and association measure.
- A numerical outcome and a categorical variable: Compare exam scores across course formats.
- Two categorical variables: Examine whether the distribution of one category differs across the categories of the other.
The appropriate display or statistical method depends on the measurement types, the study design, and relevant assumptions. A plot or summary can describe what is observed; an inferential analysis addresses what the data support beyond that description. For an example of two-population comparisons, see Penn State’s STAT 500 lesson on comparing two population parameters.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteWhat does multivariate or multivariable analysis do?
These terms are not used consistently across disciplines. In broad applied usage, “multivariate” may refer to an analysis involving several variables. In more technical usage, multivariate often means that multiple outcome variables are modeled jointly, while multivariable often describes one outcome modeled with several predictors.
Rank #3
Because the convention varies, make the model’s structure explicit: state how many outcomes it has and which variables are predictors. For example, “exam score is the outcome; study hours and course format are predictors” is clearer than relying on the word “multivariate” alone. The National Academies’ reference guide and the University of Southampton’s statistics glossary reflect distinctions and usage differences in this terminology.
Including more variables does not automatically make an analysis better. Additional variables should serve the research question, and the chosen model should fit the data and intended interpretation.
How do you choose an analysis?
- Write down the question. Are you describing one distribution, comparing groups, estimating an association, accounting for other factors, or examining multiple outcomes together?
- Identify the variables and their roles. Note which are categorical or numerical, and whether the question treats any as outcomes or predictors.
- Choose the level of analysis that fits. A distribution question may need one variable; a relationship or group comparison needs two; a question about several predictors or outcomes may call for a suitable multi-variable model.
- Check that the method fits the data and design. Measurement scale, group structure, study design, and method assumptions affect which statistical procedure is appropriate.
- Describe the result precisely. Distinguish a descriptive summary from an inferential finding, and identify the outcomes and predictors in a model.
This is a way to organize the decision, not a universal test-selection recipe. The specific method depends on details of the data and question. Curtin University’s guides explain why data and variable types and descriptive summaries matter when choosing and interpreting analyses.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Example: exam scores, study hours, and course format
Imagine a class dataset containing three variables: exam score, study hours, and course format. The sequence below shows how the same data can answer different questions; it does not report a real study result.
Best Value
- Univariate: Describe exam scores, study hours, and course format separately. For example, summarize score distributions, inspect study-hour values, and count students in each course format.
- Bivariate: Explore exam score against study hours, or compare scores across course formats. Each analysis considers a pair of variables.
- Several variables together: If the question asks how study hours and course format relate to score when considered together, model score as the outcome and those two variables as predictors. Call this multivariable or multivariate according to the convention you define.
The last model answers a different question from either pairwise exploration: it considers the predictors together. Its interpretation depends on the model and study design; it should not be described as proving that a predictor caused a score difference.
Common terminology and interpretation pitfalls
- Calling a one-variable summary a relationship: A univariate distribution describes only that variable.
- Treating “multivariate” as unambiguous: Say whether there is one outcome with multiple predictors, multiple outcomes, or both.
- Choosing a method by variable count alone: Two variables do not automatically imply one specific test; data types, design, and assumptions matter.
- Assuming a more complex model is preferable: Complexity is useful only when it helps answer the question with the available data.
- Confusing association with causation: Finding that variables vary together does not, on its own, establish that one caused the other.
Further reading
For definitions and examples of one- and two-variable analyses, see the University of West Georgia’s tutorial on univariate and bivariate analyses. For a teaching-oriented progression from univariate to bivariate analysis, see the University of Zurich’s bivariate statistics recap.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
