Finding Percentiles and Critical Values
What you'll learn: How to work backwards from a given probability to find the corresponding value in a normal distribution.
The Reverse Problem
Until now, you've been given a value (or z-score) and asked to find the probability. Now we flip it: given a probability, find the x-value or z-score that produces it. This is essential for confidence intervals, hypothesis testing, and decision-making.
How It Works
A percentile is the value below which a certain percentage of the data falls. For example, the 90th percentile is the value where 90% of the distribution lies to the left.
A critical value is a specific percentile used in statistical inference—often the boundary values for decision regions.
Finding Values from Probabilities
Given: A probability p (say, 0.95)
Find: The z-score z such that P(Z ≤ z) = 0.95
You use the standard normal table in reverse:
- Look for p in the body of the z-table (not the margins)
- Read the corresponding z-score from the row and column labels
- If needed, convert from z back to x using: x = μ + zσ
Common Critical Values
- z₀.₀₅ = 1.645 (90th percentile, for one-tailed tests)
- z₀.₀₂₅ = 1.96 (97.5th percentile, for two-tailed 95% confidence)
- z₀.₀₁ = 2.33 (99th percentile)
Analogy: Think of it like finding "what temperature gives us a 10% chance of snow?" instead of "what's the chance of snow at 32°F?" You're solving for the input, not the output.
Key Takeaway: Finding percentiles and critical values means using the z-table backwards—locating a probability inside the table and reading off the corresponding z-score, then transforming back to the original scale if needed.