Parallel Coordinates Plot Explained: How to Make, Read and Report It
A parallel coordinates plot draws every case as a line across several vertical axes, one per variable, so you can compare whole profiles on many measurements in one figure. This guide shows how to build and read one, how to scale and order the axes, how to test which variables separate your groups, and how to report the result in APA7.
What is a parallel coordinates plot?
In a scatterplot each case is a point in a space with two axes. With six or ten variables you cannot draw that space, so a parallel coordinates plot (Inselberg, 1985) lays the axes side by side as parallel vertical lines and joins each case’s values with a polyline. Each line is one case (a patient, plant or sample) and each vertical axis is a variable.
The plot is common in clinical research, genomics, ecology and engineering because it shows multivariate profiles directly: cases with a similar profile form a bundle, and unusual cases stand apart.
How to read it
- Bundles of lines that run together between two axes indicate a positive association between those two variables.
- Lines that cross in an X pattern between two axes indicate a negative association.
- A bold group mean line shows the typical profile of a group. Group means that separate on an axis mark a variable on which the groups differ.
- A line far from the rest is a possible outlier or a distinct subgroup worth checking.
Because neighbouring axes show relationships only for adjacent pairs, the order of axes matters. Ordering axes by effect size or by similarity places the most informative variables next to each other.
Scaling the axes
If variables have different units (mmHg, mg/dL, kg/m²) the raw plot is misleading. Min–max scaling stretches each variable to the same height while keeping the shape of its distribution. Standardized (z-score) scaling centres each variable on its mean and divides by its standard deviation, so slopes are comparable across variables. State which one you used in the figure note.
Test what the picture suggests: one-way ANOVA per variable
A picture is not a test. When you have a grouping column, test each variable with a one-way ANOVA and report F, degrees of freedom, p and an effect size such as eta squared (η²). Because you test many variables, adjust the p-values; the Holm procedure (Holm, 1979) controls the family-wise error rate and is more powerful than Bonferroni, and the Benjamini–Hochberg procedure (1995) controls the false discovery rate.
The Parallel Coordinates Plot Generator runs the ANOVA for every variable, shows raw and Holm-adjusted p-values and η², and orders axes by η² if you ask. If an effect is significant for a variable with three or more groups, follow up with a post hoc test such as Tukey’s HSD. For several outcomes analyzed jointly see MANOVA.
Step by step: make a parallel coordinates plot in StatFas
- Open the Parallel Coordinates Plot Generator.
- Load the example data, paste a table from Excel, or upload a CSV, Excel or SPSS file with three or more numeric variables and one optional group column.
- Pick the variables, the group to colour by, the axis scaling (min–max or standardized), the axis order (as listed, by η², or by similarity), the line opacity and whether to draw the bold group mean profile.
- Run the analysis and read the ANOVA table, the group means table and the figure with its interpretation.
- Adjust colors, fonts, line widths and sizes, then log in (free) to download the figure, tables and a Word report.
The tutorial repeats the plot in SPSS, R, Python and MATLAB.
How to report it in APA7
Name the plot and the scaling in the figure note, report one ANOVA result per variable that differs (F, degrees of freedom, adjusted p, η²), and say which correction you used. Example wording (illustrative numbers):
Common mistakes to avoid
- Plotting variables on raw, unrelated scales.
- Drawing hundreds of opaque lines so the figure becomes a solid block. Use transparency or show group means.
- Concluding a difference from the figure without a test or without adjusting for many comparisons.
- Forgetting that axis order changes which relationships are visible.
Frequently asked questions
When should I use a parallel coordinates plot instead of a scatterplot matrix?
Use it when you have more than about five variables and want to compare profiles of groups. A scatterplot matrix shows pairwise relationships but becomes crowded as variables increase.
Should I use z-scores or min–max scaling?
Use z-scores when you want comparable slopes and to see how far groups are from the mean in standard deviations. Use min–max when you want each axis to span the same visual range.
How many variables can I include?
Up to about ten to fifteen stay readable. Beyond that consider a PCA biplot or a correlogram.
Can I make a parallel coordinates plot for free?
Yes. The StatFas generator is free to run in your browser; a free account is needed only to download results.
Related tools, guides and tutorials
- Tutorial: parallel coordinates plot in StatFas, SPSS, R and Python
- One-way ANOVA calculator
- One-way ANOVA tutorial
- One-way ANOVA explained
- MANOVA explained
- Tukey’s HSD explained
- Descriptive statistics calculator
- Levene’s test explained
- Normality tests explained
- PCA Biplot Generator
- 3D Scatter Plot Generator
- Ternary Plot Generator
- All tutorials
- All blog posts
References
- American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).
- Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society B, 57(1), 289–300.
- Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6(2), 65–70.
- Inselberg, A. (1985). The plane with parallel coordinates. The Visual Computer, 1(2), 69–91.
- Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.