Course contentsShow
← Data Science
Lesson 352 of 2,14510. Hypothesis Tests for MeansFree lesson

The t-Distribution and Degrees of Freedom

How the t-distribution differs from the normal distribution and why degrees of freedom (n-1) matter for small samples.

The t-Distribution and Degrees of Freedom

What you'll learn: How the t-distribution differs from the normal distribution and why degrees of freedom matter when working with small samples.

What Is the t-Distribution?

When you estimate a population mean from a small sample, you face extra uncertainty—not only about the mean itself, but also about the population standard deviation (which you're estimating with your sample's standard deviation). The t-distribution accounts for this double-dose of uncertainty.

Think of it like this: the normal distribution is a confident expert who knows everything about the population. The t-distribution is a cautious beginner who knows there's more they don't know, so it hedges its bets with wider, heavier tails.

Key Differences from the Normal Distribution

  1. Heavier tails: The t-distribution has more probability in its tails, making extreme values more likely
  2. Lower peak: The center is slightly flatter than a normal curve
  3. Shape changes with sample size: As your sample grows, the t-distribution gradually approaches the normal distribution

Why Degrees of Freedom (n-1) Matter

Degrees of freedom represent the amount of independent information you have. When you calculate a sample standard deviation, you first compute the sample mean, which "uses up" one piece of information. If you know the mean and all but one data point, that last value is determined—it's not "free" to vary.

For a sample of size n, you have n-1 degrees of freedom for inference about the mean.

With fewer degrees of freedom (smaller samples), the t-distribution has heavier tails—reflecting greater uncertainty. As df increases, those tails thin out until the t-distribution becomes indistinguishable from the normal distribution (typically around 30+ degrees of freedom).

Key Takeaway: The t-distribution adjusts for the extra uncertainty in small samples by using heavier tails that widen with fewer degrees of freedom (n-1), providing more honest inference when you don't know the population standard deviation.