Independence and Conditional Distributions
What You'll Learn
You'll discover the elegant relationship between independence and conditional distributions: two random variables are independent if and only if conditioning doesn't change the distribution.
The Core Idea
When two random variables X and Y are independent, knowing the value of X tells you nothing about Y's distribution. Mathematically, this means:
X and Y are independent ⟺ f_{Y|X}(y|x) = f_Y(y) for all x and y where the densities are positive.
In plain English: if the conditional density of Y given X equals the marginal density of Y, then X and Y are independent. The conditional distribution is identical to the unconditional one—conditioning on X gives you no new information about Y.
This works both ways:
- If X and Y are independent, then f_{Y|X}(y|x) must equal f_Y(y)
- If f_{Y|X}(y|x) = f_Y(y) for all relevant x and y, then X and Y must be independent
Why This Matters
Think of it like weather forecasting: if today's weather (X) and tomorrow's lottery numbers (Y) are independent, then learning today is sunny tells you nothing new about tomorrow's lottery. The distribution of Y remains unchanged: f_{Y|X}(y|sunny) = f_Y(y).
But if X is today's temperature and Y is tomorrow's temperature (not independent!), then f_{Y|X}(y|90°F) ≠ f_Y(y)—conditioning changes the distribution because they're related.
Testing Independence
This gives you a practical test: compute f_{Y|X}(y|x) and compare it to f_Y(y). If they're equal everywhere, you have independence. If they differ anywhere, you don't.
Key Takeaway: Independence means conditioning provides no information—the conditional distribution equals the marginal distribution: f_{Y|X}(y|x) = f_Y(y) for all x, y with positive density.