How to explain it
At this standard, students construct scatter plots from bivariate data and interpret patterns of association by identifying direction (positive, negative, or none), form (linear or nonlinear), and features such as clustering and outliers.
The anchor students hold onto: Describe a scatter plot by its association type (positive, negative, or none) and its form (linear or nonlinear). Always check for clustering and outliers.
In Lines of Best Fit (8.SP.A.2), students use scatter plot patterns from this skill to draw and assess linear models through the data.
Worked examples
Example 1
Positive Linear
Hours practiced vs. free-throw %
Step 1Direction: rises left to right → positive.
Step 2Form: roughly straight line → linear.
Step 3Association: positive linear association.
AnswerPositive linear association.
Example 2
Negative Nonlinear
Miles (x) vs. tread depth (y)
Step 1Direction: drops left to right → negative.
Step 2Form: curve flattens at high x → nonlinear.
Step 3Association: negative nonlinear association.
AnswerNegative nonlinear association.
Example 3
No Association
Shoe size (x) vs. math score (y)
Step 1No consistent rise or fall → no trend visible.
Step 2Form: cannot be determined — no pattern.
Step 3Association: no association.
AnswerNo association.
Common mistakes
What students write
Calling a spread-out cloud "no association" when it still has a directional trend.
The fix
Spread = weak association, not no association. Look for any consistent rise or fall first.
Try this
A student analyzed the scatter plot shown. Here is their work: The y-values (9, 8, 7, 6 ..) are all positive numbers greater than 0. Since the y-values are positive, this must be a positive association. Conclusion: positive linear association. Find the error. Then write the correct association description.
What students write
Saying a scatter plot "proves" one variable causes another.
The fix
Scatter plots show association only. Causation requires controlled experiments, not just correlation.
Teacher tip
Head off the two predictable errors before they happen. First: Spread = weak association, not no association. Look for any consistent rise or fall first. Second: Scatter plots show association only. Causation requires controlled experiments, not just correlation.