Multiple regression
A regression with several explanatory factors at once. Each coefficient estimates the link between one factor and the outcome with the other factors in the model held fixed.
In plain English
It's the standard way to deal with measured confounders when you can't run an experiment. Adding a confounder can shrink, erase or flip a link. But it can only hold fixed what was measured: an unmeasured confounder still biases the result, and controlling for a step on the causal path can hide a real effect.
Uncle Gary explains
In a simulated survey where coffee has no effect but smokers drink more coffee, coffee alone seems to cost 3.5 fitness points a cup; with smoking added, the coffee effect is zero and smoking costs 15 points.
Why it matters for your money
A factor that 'explains' returns in a simple regression may vanish once size, sector or market moves are added. Ask what else the model held fixed.
Where we tested it
Related terms
All glossary terms. Educational content, not financial advice.