📊 Data Associations: Correlation & Causation
What is it and why is it useful?
Data associations help us understand relationships between two quantitative variables. We analyze if changes in one variable are related to changes in another. This is crucial for identifying trends, making predictions, and understanding real-world connections—like studying hours vs. test scores. However, a key concept is that correlation does not imply causation.
Step-by-Step Guide
- Collect Data: Gather paired data values (e.g., (x₁, y₁), (x₂, y₂)...).
- Visualize: Create a scatter plot to see the relationship's direction and form.
- Describe: Identify the direction (positive/negative), form (linear/non-linear), and strength (strong/weak).
- Calculate: Compute the correlation coefficient (r) to quantify the strength and direction.
- Interpret: Explain what the relationship means in context, being careful not to assume cause and effect.
Visual Examples
Example 1: Study Time vs. Test Score
Data: (1, 60), (2, 70), (3, 75), (4, 85)
Steps:
- Plot points: Scatter plot shows an upward trend.
- Direction: Positive (as study time increases, score increases).
- Strength: Strong (points are close to an imaginary line).
- Interpretation: There is a positive association, but we cannot say studying more causes higher scores (other factors like prior knowledge could be involved).
Example 2: Ice Cream Sales vs. Drowning Incidents
Data shows a strong positive correlation.
Interpretation: While r is high, this is a spurious correlation. A lurking variable—hot weather—causes both to increase. Ice cream sales do not cause drownings.
🚨 Common Mistakes
- Confusing Correlation & Causation: Just because two things are related doesn't mean one causes the other. Always consider lurking variables.
- Ignoring Outliers: A single outlier can significantly impact the correlation coefficient (r). Always look at the scatter plot!
- Assuming Linearity: A low r-value doesn't always mean "no relationship"—it might just be a non-linear one.
💡 Tips & Tricks
- Memory Aid: "CORRelation = CONNECTION, not CAUSE."
- r-value Guide: |r| > 0.8 = Strong, |r| ~ 0.5 = Moderate, |r| < 0.3 = Weak.
- Always "zoom out" on your scatter plot mentally. Could a third variable be influencing both?
How to Practice
- Find real-world bivariate data sets online (sports stats are great) and create your own scatter plots.
- Practice calculating the correlation coefficient (r) by hand for small data sets to understand the formula.
- Play "correlation or causation?" with news headlines. For example, "People who floss daily live longer" – is this correlation or causation?