Topic 2.4 Representing the Relationship Between Two Quantitative Variables: construct and describe scatterplots by direction, form, strength, and unusual features, in context.

What this topic is asking

The College Board (Topic 2.4) wants you to display two quantitative variables with a scatterplot and to describe it by direction, form, strength, and unusual features, always in context. This verbal description is the two-variable counterpart of Unit 1's SOCS.

The scatterplot

The axis convention matters: explanatory on $x$, response on $y$. It fixes the meaning of "positive" and "negative" direction and sets up the regression line later, whose slope and prediction read off the same axes. A point that does not follow the convention is a frequent small error that confuses everything downstream.

Describing a scatterplot

Run through four features, the two-variable analogue of SOCS:

• Direction. Positive if $y$ tends to increase as $x$ increases (points rise left to right); negative if $y$ tends to decrease (points fall). No clear trend means no association.
• Form. Is the pattern linear (a straight-line trend) or curved (for example, exponential or quadratic)?
• Strength. How tightly do the points cluster around the underlying pattern? Tight clustering is strong; a loose cloud is weak.
• Unusual features. Outliers (points far from the pattern), clusters, or distinct subgroups that suggest a lurking categorical variable.

Why describing comes before fitting

The order of operations in Unit 2 is deliberate: you describe the scatterplot first, then decide whether a line is appropriate, then fit and assess it. The reason is that the tools that follow, the correlation coefficient and the least-squares regression line, are summaries of a linear relationship. If the scatterplot is clearly curved, a correlation near zero can hide a strong non-linear relationship, and a straight-line fit will systematically miss the pattern, so you would need a transformation (Topic 2.9) instead. If there is a glaring outlier, it can distort both the correlation and the line, so you want to notice it before it quietly skews your numbers. This is why the College Board tests scatterplot description on its own: spotting non-linearity and outliers by eye is the gatekeeping judgement that decides whether the rest of the unit's machinery even applies. A student who fits a line to an obviously curved pattern without comment has missed the point of the whole sequence.

Direction, strength, and the link to correlation

Direction and strength foreshadow the correlation coefficient $r$ of the next topic, which puts a single number on exactly these two features for a linear pattern: the sign of $r$ encodes direction and its magnitude (closeness to $1$) encodes strength. But $r$ only makes sense once you have confirmed the form is roughly linear, which is the verbal judgement you make here. Form is the feature $r$ cannot capture: a strong curved relationship and a weak linear one can share the same $r$, so the scatterplot's form must be read by eye. Keeping direction, form, and strength as three separate ideas, rather than collapsing them into "strong," is what lets you describe and later model a relationship correctly, and it is precisely the distinction the exam probes when it shows a curved scatterplot with a deceptively small or large correlation.

Try this

Q1. Points on a scatterplot form a tight upward-curving arc. State the direction and form. [2 points]

• Cue. Direction is positive (rising), but the form is curved (non-linear), so a straight-line summary would be inappropriate.

Q2. Why should you describe a scatterplot's form before computing a correlation? [1 point]

• Cue. Correlation only measures linear strength; for a curved relationship a small $r$ can hide a strong association, so you must confirm linearity first.