How to interpret a scatterplot

A scatterplot places each paired observation as a point on an x-y coordinate plane. It helps you see whether two numerical variables tend to move together and whether a straight-line model may describe that pattern. Before interpreting the cloud of dots, read the axis labels, units, and scale. A plot of hours studied versus score means something different from a plot that reverses the axes, and a truncated scale can make a modest pattern look dramatic.

Describe the direction

A positive association means larger values of one variable tend to occur with larger values of the other; the points generally rise from left to right. A negative association means larger values of one variable tend to occur with smaller values of the other; the pattern generally falls from left to right. No clear upward or downward tendency suggests little or no association in the display.

Describe the form

A roughly straight band suggests a linear pattern. A curved pattern may mean a straight-line summary misses the relationship. A cluster, gap, or bend can signal that the data contain subgroups or that a single model is inappropriate. Do not force a linear interpretation just because the axes are numbered or a trend line has been drawn.

Describe the strength

In a strong association, points lie relatively close to a clear pattern. In a weak association, points are more dispersed. Strength is judged relative to the form and scale, not by whether every point follows the pattern. A single outlier can influence a calculated correlation substantially, so note unusual points before relying on a summary statistic.

Association is not causation

A scatterplot can show association, not by itself a cause-and-effect relationship. A third variable may influence both measurements; the direction of influence may be unclear; or the pattern may reflect how the data were selected. For example, ice-cream sales and swimming activity could rise together in warm months without one causing the other. The weather is a plausible common factor.

Use a line of best fit carefully

A line of best fit summarizes a linear trend and can support interpolation within the observed data range. Extrapolating beyond that range is riskier because the relationship may change. The line does not pass through every point, and a high or low value predicted from it is an estimate. If the plotted pattern is curved, a straight-line prediction may be misleading.

A short interpretation template

Name the variables and describe the pattern: “As [x] increases, [y] tends to increase/decrease; the association is roughly linear/curved and weak/moderate/strong, with [notable cluster or outlier]. This plot shows association, not proof that x causes y.” Use the exact variables and avoid stronger claims than the data support.

Common exam mistakes

• Calling a positive association “good” or a negative association “bad”; positive and negative describe direction only.
• Ignoring axis units or interval sizes.
• Treating a cluster or outlier as irrelevant without checking whether it changes the pattern.
• Assuming a line of best fit proves causation.
• Predicting far outside the observed x-values without noting that this is extrapolation.

On the FTCE, a complete response usually needs more than “the variables are related.” Identify direction and form, support the description with the displayed pattern, and keep causal language out unless the question provides evidence from a design capable of establishing causation.

Read the variables and units on both axes

A scatterplot places paired observations at coordinates (x, y). Before interpreting the point pattern, identify what each axis measures, its units, and the population represented. A point at (4, 80) means x = 4 and y = 80 according to those labels; it does not identify a participant unless the plot says so. Reversing the axes changes which variable is treated as explanatory.

Describe the overall direction, form, and strength. Direction may be positive or negative; form may be roughly linear or curved; strength reflects how closely the points follow the pattern. A graph can show a strong curved association even when a linear correlation is near zero. Inspect the point cloud rather than relying on one statistic.

Notice clusters, gaps, and outliers

Clusters may indicate subgroups with different behavior, such as separate age ranges or treatment groups. A gap may show that no observations occur in an interval. An outlier lies far from the main pattern and can influence a fitted line or correlation. Check whether it is a recording error, a valid unusual case, or evidence that the relationship differs for that observation.

A cluster can change the story of the combined data. Two groups may each show little or negative association, while their combined observations show a positive trend because the groups have different averages. When labels reveal subgroups, compare their patterns separately before describing the overall plot.

Interpret the pattern within its limits

A scatterplot shows association, not necessarily causation. A third variable may influence both measures, and the direction of influence may be reversed. A plot also cannot show how the relationship was measured or whether observations were sampled representatively unless that information is provided. State the pattern and its limitation rather than extending the conclusion.

A truncated axis can exaggerate visual differences, and unequal scales can alter apparent steepness. Use the tick labels and calculate changes from coordinates where possible. If a line of best fit is shown, describe it as a model of the average pattern; individual points need not lie on it.

• Read variables, units, sample, and axis scales first.
• Describe direction, form, and strength separately.
• Check for subgroups, gaps, and influential outliers.
• A near-zero linear correlation does not rule out a curved pattern.
• Do not infer causation from association alone.

Apply it to the evidence or sentence

A scatterplot shows paired quantitative observations. Describe direction (positive, negative, or no clear association), form (linear or curved), strength, and any clusters or outliers. A positive association means larger values of one variable tend to accompany larger values of the other; it does not mean every point follows a line. An outlier can strongly affect a correlation coefficient, so inspect the plot before summarizing the relationship with one number. Scatterplots do not establish causation, and extrapolating beyond the observed range can be unreliable. State which variables are on each axis and describe the pattern without claiming more than the data show.