Scatter Plot
A scatter plot places each measurement as a point on two axes to reveal whether two variables are related and how strongly.
Definition
A scatter plot is a graph made of dots, one for each measurement, with no lines connecting them, used to see whether two things go together. Paired measurements are plotted with one variable on each axis; points clustering along a rising or falling line suggest a correlation, and a shapeless cloud suggests none. Because every individual observation is shown as an $(x, y)$ point, the plot preserves the full spread of the data rather than collapsing it to means, making it the standard tool for judging the form, direction, and strength of a relationship and for spotting outliers and clusters before any model is fitted. A best-fit line, its equation, and a correlation coefficient can summarize the pattern, but the plot guards against summarizing data whose shape the statistic cannot capture.
Example
For each bean plant, a dot at its daily hours of sunlight (bottom axis) and its height (side axis) slopes upward if plants with more sun grew taller. Twenty students each measure a rock sample's mass and displaced volume and plot one point per sample; the points line up through the origin with slope about $2.7\ \text{g/cm}^3$, showing that mass and volume are proportional and the rocks share one density. Plotting all 15 individual cart trials (three at each of five ramp angles) instead of the five means reveals that the spread grows at steeper angles, which the mean-only line graph had hidden, so the group weights the steep-angle points less when fitting.
Key Insight
Look at the shape of the whole cloud of dots, not at any one dot; the pattern in the cloud is what the graph is for. A scatter plot is the right choice when you measured both variables rather than setting one, because it shows what the data say about the relationship before you assume one. Means summarize and scatter plots reveal: two data sets with identical means and trend lines can have completely different shapes, so always look at the raw points first.