High-Low vs Regression: Why Does Regression Usually Use More Information?
Both estimate cost behavior, but high-low uses two activity points while regression can use the broader data set.
High-low
Use the highest and lowest activity levels, not the highest and lowest costs. The change in cost divided by the change in activity estimates variable cost per unit.
High-low weakness
The method ignores all middle observations, making it simple but potentially less representative.
Regression
Regression can use many observations simultaneously. In a simple cost model, slope can represent estimated variable cost and intercept estimated fixed cost.
R-squared
R-squared describes how much variation in the dependent variable is explained by the model, but it does not prove causation or economic appropriateness.
Correlation is not causation
Variables moving together does not prove one causes the other.
The takeaway
High-low is simple and uses two points. Regression can use the broader data set.
Related Summit guides
FAQ
Does high-low use highest cost and lowest cost?
No. Use highest and lowest activity levels.
What can slope represent?
Estimated variable cost per activity unit in a simple cost model.
Does high R-squared prove causation?
No.