SPSS command: SPSS OVERALS
OVERALS nonlinear canonical correlation in your browser
OVERALS asks what two or more sets of variables have in common.
Free · No account · Files stay on your device · Updated 18 October 2026
What it tells you
Each set is combined into scores that agree as closely as possible with the scores of the other sets, while the categories of each variable get optimal numeric values. Crosstabs runs the SPSS OVERALS algorithm on your device with multiple nominal, single nominal, ordinal and numeric levels.
- The loss of each set per dimension, the eigenvalues and the total fit.
- Canonical correlations derived from the eigenvalues, which reduce to ordinary canonical correlations for two numeric sets.
- Component loadings and category quantifications that show which variables drive the shared dimensions.
When to use it
Use it when
- You have blocks of survey questions (for example attitudes and behaviours) and want the dimensions they share.
- The variables are categorical or ordinal, so linear canonical correlation would force arbitrary codes.
Use something else when
- All variables are numeric and only two sets are compared: linear canonical correlation gives the same answer with tests.
- You need significance tests or case weights: OVERALS is descriptive and every case counts once.
How to run it in Crosstabs
- 01
Open your data
Open a .sav, CSV or Excel file in the workspace and switch to Statistics mode. The file is read in your browser and never uploaded.
- 02
Choose Nonlinear canonical correlation (OVERALS)
In the Dimension reduction group, pick Nonlinear canonical correlation (OVERALS).
- 03
Put variables in sets
For each variable choose Set 1 to Set 6 and an optimal scaling level (Multiple nominal, Single nominal, Ordinal or Numeric). At least two sets are needed.
- 04
Set dimensions and run
Set Dimensions in solution and Maximum iterations, then click Run.
Worked example
Computed by the Crosstabs engine from the synthetic data below when this page was built — not typed in.
Data
The 480 respondents of the bundled "Product experience (numeric)" sample (seeded). Set 1, the experience: “The product is easy to use” and “Support resolves my issues” (both Ordinal). Set 2, loyalty and plan: “I would recommend the product” (Ordinal) and Plan (Single nominal). 2 dimensions, at most 100 iterations.
| Item | Dimension 1 | Dimension 2 | Sum |
|---|---|---|---|
| Loss, set 1 | 0.176 | 0.437 | 0.614 |
| Loss, set 2 | 0.176 | 0.437 | 0.613 |
| Mean loss | 0.176 | 0.437 | 0.613 |
| Eigenvalue | 0.824 | 0.563 | |
| Fit | 1.387 | ||
| Canonical correlation | 0.648 | 0.126 |
480 objects; 67 iterations. Canonical correlation = (K × eigenvalue − 1)/(K − 1) with K = 2 sets.
| Set | Variable | Scaling level | Multiple Fit (sum) | Single Fit (sum) | Single Loss (sum) |
|---|---|---|---|---|---|
| 1 | The product is easy to use | Ordinal | 0.756 | 0.752 | 0.004 |
| 1 | Support resolves my issues | Ordinal | 0.793 | 0.789 | 0.003 |
| 2 | I would recommend the product | Ordinal | 0.831 | 0.829 | 0.001 |
| 2 | Plan | Single nominal | 0.582 | 0.582 | 0.000 |
Sums over the two dimensions. Single loss is multiple fit minus single fit: what the rank-one restriction costs.
| Variable | Dimension 1 | Dimension 2 |
|---|---|---|
| The product is easy to use | 0.729 | 0.447 |
| Support resolves my issues | 0.793 | −0.365 |
| I would recommend the product | 0.904 | −0.070 |
| Plan | 0.235 | 0.725 |
Loadings of the single (rank-one) variables; without missing values they are correlations with the object scores.
| Category | Marginal Frequency | Quantification | Dimension 1 | Dimension 2 |
|---|---|---|---|---|
| Free | 235 | −1.021 | −0.088 | −0.774 |
| Pro | 175 | 0.989 | 0.085 | 0.750 |
| Team | 70 | 0.955 | 0.082 | 0.724 |
Optimal scaling level: Single nominal. Single category coordinates = quantification × weight per dimension.
How to read it
The two sets share a strong first dimension: canonical correlation 0.648, against 0.126 on the second. “I would recommend the product” loads 0.90 on it and the experience items 0.79 (support) and 0.73 (ease of use): satisfaction is what the experience set shares with loyalty. Plan loads 0.23 on dimension 1 and 0.73 on dimension 2, where Free (quantification −1.02) sits opposite the paid plans.
Output tables explained
- Summary of Analysis
- Loss per set and dimension, mean loss, eigenvalues, fit and canonical correlations.
- Fit
- Multiple fit, single fit and single loss for every variable.
- Component Loadings
- Projections of the quantified single variables on the object scores, with a loadings map.
- Quantifications
- Category quantifications with single and multiple category coordinates per variable.
Assumptions and notes
- A variable contributes only what is not already explained by the other variables of its set, as in multiple regression.
- An object with a missing value in a set is left out of that set only (listwise deletion per set).
- Nominal and ordinal solutions can reach a local optimum; the start is the numeric solution, as in SPSS's nested start.
How the results are validated
When every variable is multiple nominal or numeric, an independent reference solves OVERALS in closed form as an eigenproblem of the average projector (NumPy), including listwise deletion per set; eigenvalues, losses per set, loadings and object scores agree within 1e-6, and two numeric sets reproduce SciPy-based canonical correlations. Single nominal and ordinal solutions match an independent NumPy alternating least squares (best of 12 starts, confirmed as a fixed point) within 1e-6. Validation evidence
Questions
How is the canonical correlation computed?
From each dimension's eigenvalue E and the number of sets K as (K × E − 1)/(K − 1). For two numeric sets it equals the ordinary canonical correlation.
What is the difference between single and multiple nominal?
A single nominal variable gets one set of category values and a weight per dimension; a multiple nominal variable gets separate category values on every dimension.
Can I use one variable per set?
Yes. With one variable per set OVERALS becomes nonlinear principal components (PRINCALS), which also has its own procedure.
Will results match SPSS to every decimal?
Converged numeric and multiple nominal solutions are unique and should agree; nominal and ordinal solutions can differ if SPSS's random start reaches another local optimum.
Related procedures
- Nonlinear Principal Components (PRINCALS)PRINCALS
- Categorical Principal Components (CATPCA)CATPCA
- Multiple Correspondence AnalysisMULTIPLE CORRESPONDENCE
- Factor AnalysisFACTOR
- Free SPSS alternative: full coverage list
Run Nonlinear Canonical Correlation (OVERALS) on your own data
Open a .sav, CSV or Excel file. Everything runs in your browser; the file never leaves your device.
SPSS is a trademark of IBM Corp. Crosstabs is not affiliated with or endorsed by IBM.