Interpreting Dummy Variable Coefficients
What you'll learn: How to read dummy variable coefficients as the difference in outcome compared to the reference category.
The Core Idea
When you include a dummy variable in your regression model, its coefficient tells you how much higher (or lower) the outcome is for that category compared to the reference category, holding other variables constant.
Think of it like comparing heights across groups. If your reference category is "children" and you have a dummy for "adults," a coefficient of +15 would mean adults are 15 units taller on average than children.
Reading the Coefficient
The coefficient of a dummy variable represents:
- The average difference in the outcome (Y) between observations in that category and observations in the reference category
- This difference is expressed in the same units as your outcome variable
- It's interpreted ceteris paribus (all other predictors held constant)
Example
Suppose you're predicting salary with a dummy variable Female (reference = Male):
- If the coefficient for
Femaleis -5000, this means women earn $5,000 less on average than men (the reference group), holding other factors constant - If the coefficient were +2000, women would earn $2,000 more than men on average
Sign and Magnitude Matter
- Positive coefficient → that category has a higher average outcome than the reference
- Negative coefficient → that category has a lower average outcome than the reference
- Magnitude → the size of the difference in outcome units
Key Takeaway: A dummy variable's coefficient tells you the average difference in outcome between that category and the reference category—it's always interpreted relative to what you left out.