Multivariate analysis is one of the most powerful sets of techniques in quantitative research. It allows you to examine relationships between multiple variables simultaneously – giving you a richer, more complete picture than single-variable tests ever could.
However, many researchers hesitate when they need to know how to perform multivariate analysis in SPSS. The range of available techniques – MANOVA, multivariate regression, factor analysis, discriminant analysis – can feel overwhelming without a clear starting point.
Therefore, this guide breaks the process down step by step. You will learn what multivariate analysis is, which technique to choose for your research question, how to run each method in SPSS, and how to interpret the output correctly.
Whether you are working on a dissertation, academic study, or professional research project, this guide gives you everything you need to get started with confidence.
What Is Multivariate Analysis?
Multivariate analysis refers to a collection of statistical techniques that analyse more than one outcome variable at the same time. Rather than testing each dependent variable in isolation, these methods examine the combined effect across all variables together.
As the name implies, multivariate regression is a technique that estimates a single regression model with more than one outcome variable. When there is more than one predictor variable in a multivariate regression model, the model is a multivariate multiple regression.
This approach offers two key advantages over running separate univariate tests. First, it controls the experiment-wide error rate by avoiding the inflation of Type I error that occurs when you run many individual tests. Second, it captures the relationships between outcome variables – something that separate analyses miss entirely.
The most commonly used multivariate methods in SPSS include:
- MANOVA (Multivariate Analysis of Variance) – tests group differences across multiple dependent variables simultaneously
- Multivariate Regression – estimates coefficients for multiple outcome variables using one set of predictors
- Factor Analysis – reduces many variables into a smaller number of underlying factors
- Discriminant Analysis – classifies cases based on a set of predictor variables
- Cluster Analysis – groups cases into homogeneous clusters based on similarity across variables
Each technique serves a different research purpose. Choosing the right one depends entirely on your research question, your variable types, and your data structure. Before conducting any analysis, always ensure your SPSS data collection process produces clean, complete, and properly formatted data.
Turn Survey Data Into Business Decisions Faster
Technology-driven market research for faster, smarter insights.
When Should You Use Multivariate Analysis?
Use multivariate analysis when your research involves two or more dependent or outcome variables that you want to analyse together. Analysing them separately misses the correlations between them and produces incomplete conclusions.
Here are typical scenarios where learning how to perform multivariate analysis in SPSS is essential:
- You want to test whether a teaching method affects scores in both Math and Science simultaneously
- You are studying how psychological variables (motivation, self-concept) relate to a set of academic outcomes
- You want to reduce a long survey into a smaller number of meaningful underlying dimensions
- You need to classify patients into diagnostic groups based on multiple clinical measurements
The outcome variables should be at least moderately correlated for the multivariate regression analysis to make sense. Multivariate regression analysis is also not recommended for small samples.
In addition, always verify that your dataset is complete before running any multivariate test. Read our full guide on how to delete missing data in SPSS to prepare your dataset properly. Missing values in any variable can distort results significantly.
Key Assumptions of Multivariate Analysis in SPSS
Before you perform multivariate analysis in SPSS, you must verify several critical assumptions. Violating them compromises the validity of your results.
The key assumptions underpinning multivariate analysis include: multivariate normality – the dependent variables should collectively exhibit a multivariate normal distribution; homogeneity of covariance matrices – the covariance matrices of the dependent variables across different groups should be roughly equal; and linearity – the relationships between the independent variable and each dependent variable should be linear.
In practice, check these assumptions before running any test:
- Multivariate normality – Run Mahalanobis distance via regression diagnostics to check for multivariate outliers
- Homogeneity of covariance matrices – Box’s M test in SPSS checks this directly (available in the MANOVA options)
- Linearity – Inspect scatterplot matrices of all dependent variables with your predictors
- No multicollinearity – Run correlation checks between your predictor variables. High correlations (above 0.90) between predictors indicate a problem
- Adequate sample size – A general rule is 10–20 cases per variable, minimum
Skipping assumption checks is the most common reason multivariate results are rejected in peer review. Therefore, always document your assumption testing in your methodology section.
Method 1: How to Perform MANOVA in SPSS (Step by Step)
MANOVA (Multivariate Analysis of Variance) is the most commonly used technique when learning how to perform multivariate analysis in SPSS. It tests whether group membership on a categorical independent variable significantly predicts differences across a combination of dependent variables.
Step 1 – Load Your Dataset
Open SPSS and load your dataset. Go to File > Open > Data and select your .sav or Excel file.
Step 2 – Access the MANOVA Menu
In the top menu, click on Analyze. Within the Analyze menu, navigate to General Linear Model and choose Multivariate. This is accessed via Analyze > General Linear Model > Multivariate.
Step 3 – Specify Variables
In the Multivariate dialogue box:
- Move your dependent variables (e.g. Math scores, Science scores) into the Dependent Variables box
- Move your independent variable (e.g. teaching method) into the Fixed Factor(s) box
Step 4 – Set Options
Click the Options button and check Descriptive statistics, Homogeneity Tests, and Estimates of effect size. These options produce the additional diagnostics and effect size information needed for a complete interpretation.
Step 5 – Add Post-Hoc Tests and Plots
Click Post Hoc, move your factor into the box, and select Tukey (for equal variances) or Games-Howell (for unequal variances). Click Continue.
Click Plot, move your factor to the Horizontal Axis, click Add, then Continue.
Step 6 – Run the Analysis
Click OK. SPSS generates a comprehensive output window with multiple tables covering descriptives, Box’s M, Multivariate Tests, Levene’s Test, and Between-Subjects Effects.
For hands-on practice before running your actual research data, download a practice dataset for SPSS and work through a MANOVA example from start to finish.
Method 2: How to Perform Multivariate Regression in SPSS
Multivariate regression extends standard regression by allowing multiple outcome variables to be predicted simultaneously from a single set of predictors. In SPSS, you run this using either the GLM or MANOVA command.
To conduct a multivariate regression in SPSS, you can use either of two commands – GLM or MANOVA. Using the lmatrix subcommand in the GLM command, you can test if all of the equations, taken together, are statistically significant.
Here is the syntax for running multivariate regression using GLM:
GLM locus_of_control self_concept motivation WITH read write science
/PRINT = PARAMETERS
/LMATRIX ‘multivariate test of entire model’ write 1; read 1; science 1.
Alternatively, use the MANOVA command:
MANOVA locus_of_control self_concept motivation WITH read write science
/PRINT PARAMETERS.
If you ran a separate OLS regression for each outcome variable, you would get exactly the same coefficients, standard errors, t- and p-values, and confidence intervals. The key advantage of multivariate regression is that you can conduct tests of the coefficients across the different outcome variables – something separate regressions cannot do.
Understanding how regression relates to multivariate tests is essential. For a thorough foundation, read our detailed guide on correlation vs regression analysis before running your first multivariate regression model.
Interpreting the MANOVA Output
Turn Complex Datasets Into Strategic Business Growth
Enterprise-grade data processing, statistical analysis, and customized tabulations to power your insights.
Once you run the analysis, SPSS produces several output tables. Understanding each one is essential for a valid interpretation.
Descriptives Table
This shows the mean, standard deviation, and N for each dependent variable across each group level. Review this first to understand your data before examining significance.
Evaluate the means and standard deviations of each group to get an initial overview of central tendency and variability. Confirm the number of observations in each group, as discrepancies in sample sizes could impact the interpretation.
Box’s M Test
This tests the assumption of homogeneity of covariance matrices. A non-significant result (p > 0.05) confirms that the assumption is met. A significant Box’s M with a large sample is not necessarily a serious violation, as the test is sensitive to non-normality.
Multivariate Tests Table
This is the core of your MANOVA output. SPSS produces four multivariate test statistics:
| Test Statistic | When to Use |
| Wilks’ Lambda | Most commonly reported; suitable for most designs |
| Pillai’s Trace | More robust when assumptions are violated |
| Hotelling’s Trace | Use when there are only two groups |
| Roy’s Largest Root | Sensitive to violations; use with caution |
A significant result on the Wilks’ Lambda test indicates that there are significant differences in the multivariate pattern of the dependent variables across the levels of the independent variable.
Report Wilks’ Lambda as your primary test statistic in academic writing, along with the F-ratio, degrees of freedom, and p-value.
Levene’s Test of Homogeneity of Variances
This tests the equality of variances for each dependent variable separately. A non-significant result (p > 0.05) for each variable confirms that the univariate variance assumption is met.
Tests of Between-Subjects Effects
This table functions like a standard ANOVA table, but is run separately for each dependent variable. It shows whether the independent variable has a significant effect on each outcome individually.
In the Between-Subjects Effects table, focus on the F-ratio. A higher F-ratio indicates larger differences among group means relative to within-group variability. Examine the p-value associated with the F-ratio – if the p-value is below 0.05, at least one group’s mean is significantly different. If available, consider effect size measures such as eta-squared, which indicates the proportion of variance explained by group differences.
Post-Hoc Tests
If your MANOVA result is significant and your independent variable has three or more levels, post-hoc tests tell you which specific groups differ. Tukey’s HSD is the standard choice when group variances are equal.
For a deeper understanding of how these statistical tests connect, explore our guide on what is correlation analysis in statistics, which explains the foundational relationships between variables that multivariate tests build upon.
Interpreting Multivariate Regression Output
The multivariate regression output in SPSS contains several key tables. Here is how to read them systematically:
Overall Multivariate Test Table:
This shows Pillai’s Trace, Wilks’ Lambda, Hotelling’s Trace, and Roy’s Largest Root for the full model. All four statistics should be checked. If all p-values are below 0.05, the overall model is statistically significant.
The tests for the overall model indicate that the model is statistically significant regardless of the type of multivariate criteria used – meaning all p-values point to the same conclusion. Below the overall model tests are the between-subjects tests, which show which predictor variables are statistically significant for each dependent variable individually.
Parameter Estimates Table:
This shows the regression coefficient (B), standard error, t-value, and p-value for each predictor for each dependent variable. Use this table to understand the direction and strength of each relationship.
Univariate F-Tests Table:
This breaks down the model significance variable by variable. It shows which dependent variables are significantly predicted by the full set of predictors.
How to Perform Factor Analysis in SPSS
Factor analysis is another major technique under how to perform multivariate analysis in SPSS. It reduces a large set of variables into a smaller number of underlying dimensions called factors.
Go to Analyze > Dimension Reduction > Factor. Move all your scale variables into the Variables box. Then:
- Click Extraction – select Principal Components and set the eigenvalue threshold to 1.0
- Click Rotation – select Varimax for orthogonal rotation
- Click Options – check Suppress small coefficients (below 0.30) for a cleaner output
- Click OK
The output shows a Component Matrix (or Rotated Component Matrix) with factor loadings for each variable. Loadings above 0.50 are considered meaningful. Variables cluster onto factors based on shared variance.
For a step-by-step walkthrough of factor analysis, see our complete guide on how to run factor analysis in SPSS. It covers extraction methods, rotation options, and output interpretation in full detail.
How to Perform Discriminant Analysis in SPSS
Discriminant analysis classifies cases into predefined groups using a set of continuous predictor variables. It is the multivariate equivalent of logistic regression for categorical outcomes.
Go to Analyze > Classify > Discriminant. Move your grouping variable into the Grouping Variable box and define its range. Move your predictor variables into the Independents box.
Click Statistics and check Means, Univariate ANOVAs, and Box’s M. Click Classify and check Summary table and Leave-one-out classification for validation.
The output produces:
- Wilks’ Lambda – tests whether group means differ significantly across the discriminant function
- Standardised Discriminant Function Coefficients – show which predictors contribute most to group separation
- Classification Results – show the percentage of cases correctly classified into each group
For a complete guide to running discriminant analysis step by step, see our article on discriminant analysis in SPSS.
Reporting Multivariate Analysis Results in APA Style
Program Complex Questionnaires and Skip Logic
Expert survey scripting, advanced routing, and multi-language configurations for flawless data collections.
When you perform multivariate analysis in SPSS and write up your findings, always follow APA reporting conventions.
For MANOVA, report:
“A one-way MANOVA was conducted to examine the effect of [independent variable] on [dependent variables]. The analysis revealed a significant multivariate effect, Wilks’ Λ = [value], F(df1, df2) = [value], p = [value], partial η² = [value].”
For Multivariate Regression, report the overall model test first, then the individual between-subjects results for each dependent variable, including B coefficients, standard errors, t-values, and p-values.
If post-hoc tests were conducted, report significant differences between specific groups. For instance, Post-hoc tests using Tukey indicated a significant difference between specific groups (p < .05). If relevant, report effect size measures such as partial eta-squared to provide insights into the practical significance of the findings.
For well-structured reporting of your complete analysis, read our guide on how to make a data analysis report. Clear, structured reporting of multivariate results is essential for academic submission and peer review.
Common Mistakes When Performing Multivariate Analysis in SPSS
Avoid these frequent errors that compromise multivariate results:
- Ignoring assumptions – Running MANOVA without checking Box’s M or Levene’s test invalidates the entire analysis
- Using too many dependent variables – More DVs require proportionally larger samples. Aim for at least 10–20 cases per DV
- Treating highly correlated DVs as independent – Check inter-correlations between all dependent variables before running MANOVA
- Reporting only univariate results – The multivariate test must be reported first. Univariate follow-ups are secondary
- Skipping post-hoc tests – A significant MANOVA tells you something differs; post-hoc tests tell you what and where
- Confusing MANOVA with MANCOVA – If you have continuous covariates to control for, use MANCOVA, not MANOVA
Quick Checklist: How to Perform Multivariate Analysis in SPSS
Before running any multivariate test, confirm each of the following:
- Research question involves two or more dependent variables
- Sample size is adequate (minimum 10–20 cases per variable)
- Data is clean with no missing values on key variables
- Multivariate normality has been visually assessed
- Box’s M or homogeneity of variance tests are planned
- Correct technique selected (MANOVA, regression, factor, discriminant)
- Effect sizes will be reported alongside significance values
- Post-hoc tests are planned if the main effect is significant
Conclusion
Learning how to perform multivariate analysis in SPSS unlocks a level of statistical depth that transforms research quality. Whether you use MANOVA, multivariate regression, factor analysis, or discriminant analysis, the process always follows the same logic – define your research question, verify assumptions, run the analysis, and interpret the output systematically.
Moreover, multivariate methods provide a more accurate picture of complex real-world relationships than any single-variable test can offer. They capture the connections between outcome variables and give you a richer, more nuanced understanding of your data.
Therefore, invest the time to master each technique step by step. Start with MANOVA as your foundation. Then expand to regression, factor analysis, and discriminant analysis as your research demands. Ultimately, competence in multivariate analysis in SPSS is one of the most valuable skills any quantitative researcher can develop.
Frequently Asked Questions
ANOVA tests the effect of one independent variable on a single dependent variable. MANOVA extends this by testing the effect of one or more independent variables on two or more dependent variables simultaneously. MANOVA also controls the experiment-wide error rate and captures the correlation structure between outcome variables.
You need at least two dependent variables. There is no strict upper limit, but each additional dependent variable requires more cases. A practical rule is to have at least 10–20 participants per dependent variable to ensure reliable results.
Wilks’ Lambda measures how much variance in the combined dependent variables is not explained by the independent variable. It ranges from 0 to 1. A value closer to 0 indicates a stronger group separation. A significant p-value (below 0.05) means the multivariate effect is statistically significant.
Yes. MANOVA, factor analysis, and discriminant analysis are all available through the graphical menu under Analyze > General Linear Model and Analyze > Classify. Multivariate regression can also be run via the GLM menu. However, using syntax gives you more precise control and a reproducible record of your analysis.
A significant Box’s M test suggests a violation of the homogeneity of covariance matrices assumption. In this case, report Pillai’s Trace instead of Wilks’ Lambda, as Pillai’s Trace is more robust to this violation. Also increase your sample size if possible, as Box’s M is highly sensitive in large samples.
