Course contentsShow
← Mathematics
Lesson 2924 of 3,12157. Common DistributionsFree lesson

Properties and Uses of the t-Distribution

Explore the mean, variance, and heavy tails of the t-distribution, and its role in small-sample inference.

Properties and Uses of the t-Distribution

You'll learn how the t-distribution's unique properties—especially its heavier tails and relationship to sample size—make it essential for real-world statistical inference when working with small samples.

Core Properties

The t-distribution looks like a normal distribution but with thicker tails, meaning it places more probability on extreme values. Here are its key characteristics:

Mean and Variance

  • Mean: 0 (when degrees of freedom ν > 1)
  • Variance: ν/(ν - 2) for ν > 2

Notice something important: the variance depends on degrees of freedom and is larger than 1 (the standard normal's variance). For example, with ν = 10, variance = 10/8 = 1.25. As ν increases, the variance approaches 1, and the t-distribution converges to the standard normal.

Heavy Tails

The t-distribution's "heavy tails" mean extreme values occur more frequently than under a normal distribution. Think of it as a safety net: when you're uncertain about population parameters (especially with small samples), the wider tails account for that extra uncertainty by making confidence intervals appropriately wider.

Why We Use It

The t-distribution appears naturally when estimating means from small samples where the population standard deviation is unknown. Instead of using the normal distribution (which assumes we know σ), we use the sample standard deviation s, which introduces additional variability. The t-distribution accounts for this extra uncertainty.

As sample size grows, degrees of freedom increase, and the t-distribution becomes nearly indistinguishable from the normal—reflecting our increased confidence with more data.

Key Takeaway: The t-distribution's heavier tails and larger variance (compared to normal) make it the correct choice for small-sample inference, providing appropriately conservative confidence intervals and hypothesis tests when population variance is unknown.