StatFAS · Q–Q Plot Calculator

Q–Q Plot Explained: Meaning, Interpretation and How to Make One in SPSS, R and Python

By the StatFAS team · Published 2026-11-07 · 7 min read

A Q–Q plot is the quickest visual check of whether a variable follows a normal distribution. This guide explains what the Q–Q plot shows, when a Q–Q plot is worth drawing, how to interpret every pattern, and how to produce a publication-ready Q–Q plot in SPSS, R, Python or with a free calculator.

Make a Q–Q plot in seconds
Paste your data, get a publication-ready Q–Q plot with Shapiro–Wilk, Anderson–Darling and APA 7 text.
Open the free Q–Q plot calculator

What is a Q–Q plot?

A Q–Q plot (quantile–quantile plot) compares the quantiles of a sample with the quantiles of a theoretical distribution, most often the normal distribution. Each dot pairs one observed value with the value that a perfectly normal sample would have at the same rank. When the dots fall along the straight reference line, the sample matches the theoretical shape. When the dots bend away from the line, the Q–Q plot reveals exactly how the data differ: skewness, heavy tails, outliers or rounding.

A normal Q–Q plot is also called a normal probability plot. A related display, the P–P plot, compares cumulative probabilities instead of quantiles and is less sensitive in the tails, which is why the Q–Q plot is preferred for normality checks.

When should you use a Q–Q plot?

  • Before a t test, ANOVA or Pearson correlation. These methods assume approximately normal data or residuals, and the Q–Q plot is the standard diagnostic.
  • After fitting a regression or ANOVA model. Draw the Q–Q plot of the residuals, not of the raw outcome.
  • When comparing a transformation. Draw one Q–Q plot before and one after a log or square-root transformation to judge improvement.
  • When a normality test disagrees with intuition. With large samples, Shapiro–Wilk flags trivial departures; the Q–Q plot shows whether the departure matters.
  • In simulation and quality work to compare a sample with exponential, uniform or other distributions.

Why a Q–Q plot beats a histogram or a p value alone

Histograms depend on bin width and hide tail behaviour. A p value depends on sample size: with 2,000 observations a harmless wobble becomes significant, and with 10 observations a serious departure goes undetected. The Q–Q plot is independent of bins, shows where the departure occurs and lets the reader judge practical importance. Best practice is to report the Q–Q plot together with the Shapiro–Wilk test and the skewness and kurtosis values.

How to interpret a Q–Q plot

Pattern in the Q–Q plotWhat the pattern means
Dots hug the straight lineData are approximately normal
Curve bends upward at the right end (convex)Positive (right) skew
Curve bends downward at the right end (concave)Negative (left) skew
S-shape, both ends leave the line outwardHeavy tails (leptokurtic), more extreme values than normal
Reverse S-shape, ends stay inside the lineLight tails (platykurtic), flatter than normal
One or two dots far from the line at an endPossible outliers
Staircase of horizontal stepsRounded or discrete data

Read the middle of the Q–Q plot for the overall fit and the ends for tail behaviour. A shaded 95% confidence envelope helps: dots inside the band are consistent with normality, and only dots clearly outside the band deserve concern. Small samples (under 30) wander naturally, so judge the Q–Q plot generously.

Common mistakes when using a Q–Q plot

  • Drawing the Q–Q plot for the raw outcome when the model assumption concerns residuals.
  • Pooling several groups in one Q–Q plot; group differences in means create a false bend.
  • Calling the sample “proven normal” because Shapiro–Wilk is non-significant; the Q–Q plot and test only show no evidence against normality.
  • Over-reading a Q–Q plot with fewer than 15 observations.
  • Ignoring severe tail departures because a large-sample test is “robust”.

How to report a Q–Q plot in APA 7

Example: Normality was evaluated using the Shapiro–Wilk test and a normal Q–Q plot (Figure 1). The Shapiro–Wilk test did not indicate a significant departure from normality, W(60) = .994, p = .992, and the observed quantiles followed the reference line closely. Report W, the sample size, the exact p value, and refer to the figure by number.

How to make a Q–Q plot in SPSS, R and Python

SPSS

  1. Open your data and choose Analyze → Descriptive Statistics → Explore.
  2. Move the numeric variable into Dependent List.
  3. Click Plots…, tick Normality plots with tests, then Continue → OK.

The output contains the Tests of Normality table (Shapiro–Wilk and Kolmogorov–Smirnov) and the Normal Q-Q Plot with a detrended version.

R

R
qqnorm(x); qqline(x, lty = 2)
shapiro.test(x)

Python

Python
import scipy.stats as st, matplotlib.pyplot as plt
st.probplot(x, dist="norm", plot=plt)
print(st.shapiro(x))
plt.show()

Complete scripts with output explanations are in the step-by-step Q–Q plot tutorial.

The easy way: Q–Q plot on StatFAS.com

The free Q–Q plot calculator runs in the browser, so data never leave the computer. The calculator draws the Q–Q plot with a quartile reference line and 95% envelope, reports Shapiro–Wilk, Anderson–Darling, Jarque–Bera, skewness and kurtosis, and writes the interpretation in APA 7 wording. Every colour, font, line width and label can be edited, and a signed-in visitor can download the whole result as one Word report.

Make a Q–Q plot in seconds
Paste your data, get a publication-ready Q–Q plot with Shapiro–Wilk, Anderson–Darling and APA 7 text.
Open the free Q–Q plot calculator

Frequently asked questions about the Q–Q plot

What does a Q–Q plot show?

A Q–Q plot shows how closely the quantiles of a sample match those of a theoretical distribution, usually the normal distribution. Dots on the straight line indicate a good match.

How do you know if a Q–Q plot is normal?

The dots form an approximately straight line, stay within the confidence envelope, and the ends do not curve away. Confirm with the Shapiro–Wilk test.

What is the difference between a Q–Q plot and a histogram?

A histogram groups values into bins, while a Q–Q plot compares every observation with the expected normal value, so the Q–Q plot shows tail behaviour and outliers more clearly.

Should I check normality of the raw data or the residuals?

For ANOVA and regression, check the residuals. For a one-sample or paired t test, check the sample or the paired differences.

How many observations does a Q–Q plot need?

A Q–Q plot works from about 8 observations, but judgments are only reliable with 20 or more.