Credible Intervals vs. Confidence Intervals
What you'll learn: You'll understand how Bayesian credible intervals differ fundamentally from frequentist confidence intervals in their interpretation and meaning.
The Core Distinction
Both intervals provide ranges of plausible parameter values, but they mean completely different things:
A confidence interval (frequentist) is about the procedure, not the specific interval you calculated. A 95% confidence interval means: "If we repeated this experiment infinitely and computed intervals each time, 95% of those intervals would contain the true parameter." Once you've computed your interval, the true parameter is either inside it or not—there's no probability about it.
A credible interval (Bayesian) is about the parameter itself, given the data you observed. A 95% credible interval means: "Given our data and prior beliefs, there's a 95% probability that the parameter lies in this interval." This is a direct probability statement about where the parameter is.
Analogy
Think of searching for buried treasure:
- Confidence interval: "My metal detector method finds the treasure in 95% of all possible treasure hunts." But on this hunt, either the treasure is in the zone I marked or it isn't—no probability.
- Credible interval: "Based on my map and today's digging clues, I'm 95% sure the treasure is in this zone." It's a probability about this specific treasure's location.
Why It Matters
The credible interval's interpretation often matches what people think confidence intervals mean. Bayesian intervals directly quantify our uncertainty about the parameter's location given the data we have. Frequentist intervals describe long-run procedural properties, which is philosophically different but equally valid under its framework.
Key Takeaway: Credible intervals give probabilities about where the parameter is (Bayesian posterior uncertainty); confidence intervals describe how often the procedure captures the parameter (frequentist long-run coverage).