Association Is a Starting Point

The concept

Association means two measured variables vary together in a pattern. A scatter plot can reveal direction, curvature, unusual points, and groups that a single summary hides. An observed relationship may result from a direct effect, reverse direction, a common cause, or selection. Causal claims require additional design and evidence.

Worked example

A fictional town finds that outdoor pool visits and sales of cold drinks rise on the same days. It would be premature to conclude that pool visits cause drink sales everywhere. Hot weather can increase both. A useful analysis records temperature, checks timing, and states the competing explanations. The graph remains informative even when the causal story is unresolved; it can guide a better study or an operational forecast.

Limits and judgment

Controlling statistically for one measured factor does not prove all confounding has disappeared. Measurements may be incomplete, relationships may be nonlinear, and combining different neighborhoods can hide patterns within each. Distinguish prediction from intervention: a variable can help forecast demand without being a useful lever to change it. The strength of the conclusion should match the study design.

RETRIEVE. REASON. REMEMBER.

Knowledge check.

Choose one answer for each question, then check your reasoning.