3D Scatter Plot Explained: How to Make, Rotate and Report It
A 3D scatter plot shows how three continuous variables relate at the same time. It is useful for exploring, but a rotated picture is not evidence by itself. This guide explains how to read a 3D scatter plot, how to back it with a regression plane, correlations and variance inflation factors, and how to report it in APA7.
What is a 3D scatter plot?
A 3D scatter plot places each case at the position given by three numeric variables, X, Y and Z. It extends the ordinary scatterplot by one dimension and lets you look for clusters, trends and outliers that depend on the combination of three variables, for example leaf area, chlorophyll and biomass, or dose, weight and response.
Because a 3D figure is projected onto a flat screen or page, depth is ambiguous. That is why interactive rotation matters on screen, and why a figure for a paper should be paired with statistics.
How to read it
- Rotate the figure until the cloud is thin; that is the direction in which the three variables are most nearly in a plane.
- A tilted band of points suggests that Z changes with X and Y together.
- Clusters by colour suggest group differences in the joint distribution.
- Isolated points are possible outliers; check them from several angles before removing anything.
Add the regression plane
The least-squares regression plane is the three-dimensional counterpart of a regression line. It shows the predicted Z for every combination of X and Y. Points above the plane have higher Z than predicted. The 3D Scatter Plot Generator fits the plane Z = b0 + b1X + b2Y and draws it on the figure.
The plane summarizes multiple linear regression, so report the same statistics: R², F with degrees of freedom and p, and for each predictor the unstandardized coefficient b with its 95% confidence interval, the standardized coefficient β, t and p.
Correlations and multicollinearity (VIF)
Start with the three Pearson correlations (see Pearson correlation explained). If X and Y are strongly correlated with each other, their separate effects on Z are hard to separate. The variance inflation factor (VIF) quantifies this; values above 5, and especially above 10, suggest that the predictors overlap too much (O’Brien, 2007). The generator reports VIF next to the regression table.
Step by step: make a 3D scatter plot in StatFas
- Open the 3D Scatter Plot Generator.
- Load the example data, paste a table, or upload a CSV, Excel or SPSS file with three numeric columns and an optional group column.
- Choose which variable is X, Y and Z (Z is the regression outcome), whether to colour by group, and whether to draw the regression plane and drop lines. Azimuth and elevation can also be typed in.
- Drag the figure to rotate it, then run the analysis for correlations, regression, VIF and the APA7 interpretation.
- Set colors, fonts, sizes and line widths, then log in (free) to download the figure, tables and Word report.
The tutorial shows the same plot in SPSS, R, Python and MATLAB.
How to report it in APA7
Include the figure with a note on the viewing angle, and report the regression in text or a table. Example wording (illustrative numbers):
Common mistakes to avoid
- Judging a relationship from one viewing angle.
- Using a 3D plot when two 2D plots would be clearer, for example for a journal figure.
- Forgetting that the regression plane assumes a linear relationship and independent observations.
- Ignoring high VIF values.
Frequently asked questions
Is a 3D scatter plot better than three 2D scatterplots?
It shows the joint pattern in one view, but depth is ambiguous on paper. Many authors show a 3D plot for exploration and 2D plots or a regression table for the final paper.
Can I rotate the plot and still download it?
Yes. Rotate to the angle you want in the StatFas generator and download the figure at that angle.
What does VIF mean?
The variance inflation factor shows how much the variance of a regression coefficient is inflated by correlation with the other predictors. Values near 1 mean no overlap.
What if my variables are not linearly related?
Consider a polynomial or non-linear model, transform the variables, or use a rank-based correlation. See the regression tools.
Related tools, guides and tutorials
- Tutorial: 3D scatter plot in StatFas, SPSS, R and Python
- Multiple linear regression calculator
- Regression calculator
- Regression tutorial
- Correlation calculator
- Pearson correlation calculator
- Pearson correlation explained
- Canonical correlation analysis
- Correlation tutorial
- PCA Biplot Generator
- Parallel Coordinates Plot Generator
- Ternary Plot Generator
- All tutorials
- All blog posts
References
- American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).
- O’Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality & Quantity, 41(5), 673–690.
- Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Lawrence Erlbaum Associates.