Scatter Plot: Visualize the Relationship Between Two Variables
How the Scatter Plot Works
A scatter plot draws each pair (X, Y) as a point, showing the relationship between two variables at a glance: positive correlation (points run up to the right), negative correlation (down to the right) or no correlation (random spread). The tool also computes the Pearson correlation coefficient r and a regression fitted line, quantifying the strength of the linear association, and is the natural first step in exploring variable relationships. Combining the graph with numbers makes correlation judgment both intuitive and quantitative.
When to Use It
Use it in quality improvement to hunt for potential causes (temperature versus strength, speed versus defect rate), to verify a hypothesized two-variable relationship, as an exploratory check before regression analysis, and to show a team the evidence of association. The scatter plot is one of the seven basic quality tools and is ideal for fast, visual communication in the early stage of a project, pointing out variable pairs worth investigating further before you design experiments to confirm causality.
How to Read the Results
Look first at the overall direction and spread of the points: r near plus or minus 1 means a strong linear relationship and near 0 means weak; the fitted line slope shows how fast Y changes with X. Watch for points far from the main body (influential points) that can distort the coefficient; the tool marks and flags suspicious points and can add an AI interpretation to separate real correlation from artifacts caused by outliers. Check significance with the p-value so small or large samples do not mislead you.
Precautions
Correlation is not causation; Pearson measures only linear relationships, so r can be near 0 for a curved relationship, and you should look at the graph before trusting r. Outliers can raise or lower the coefficient, so screen for them first, and with small samples r is unstable, so judge significance with the p-value. When needed, transform the data before assessing correlation, and report the sample size and significance level with your conclusions.