Units and Scale in Interpretation
What you'll learn: How the units of your variables X and Y shape what your regression slope actually means in the real world.
Why Units Matter
When you calculate a regression slope (β₁), you're not just getting a number—you're getting a number with units attached. The slope always has units of Y-units per X-unit, and this determines how you communicate your findings.
Understanding the Units
If you regress house price (in dollars) against square footage (in sq ft), a slope of β₁ = 150 means:
"For every additional square foot, the predicted price increases by $150."
The slope's units are dollars per square foot ($/sq ft).
Scale Changes Everything
Here's where it gets important: if you measure the same house price in thousands of dollars instead, your slope becomes β₁ = 0.15. Same relationship, different number! Now it's:
"For every additional square foot, the predicted price increases by 150)."
Practical Implications
- Always state units when reporting slopes ("$150 per square foot," not just "150")
- Rescaling changes the coefficient but not the relationship (measuring weight in kg vs. pounds changes the slope, not the actual prediction)
- The intercept's units are always the same as Y's units
- Comparing models requires matching units (you can't directly compare slopes from different scales)
Key Takeaway: The regression slope is meaningless without its units; always interpret β₁ as "change in Y-units per one X-unit" and communicate accordingly to avoid confusion.