Course contentsShow
← Mathematics
Lesson 3021 of 3,12159. Statistics — Descriptive and InferentialFree lesson

Null and Alternative Hypotheses

Formulating H₀ and Hₐ correctly for different research questions, including one-tailed and two-tailed tests.

Null and Alternative Hypotheses

What you'll learn: How to correctly state the null hypothesis (H₀) and alternative hypothesis (Hₐ) for different research questions, and distinguish between one-tailed and two-tailed tests.

The Core Idea

In hypothesis testing, we're essentially putting a claim on trial. The null hypothesis (H₀) represents the "default" or "status quo" position—typically a statement of no effect, no difference, or no change. The alternative hypothesis (Hₐ) is what we suspect might be true instead—it's the research claim we're trying to find evidence for.

Think of H₀ as "innocent until proven guilty." We assume it's true unless the data provides strong enough evidence to reject it in favor of Hₐ.

Formulating Hypotheses

Two-Tailed Tests

When you want to detect any difference (in either direction):

  • H₀: μ = μ₀ (the population mean equals some specific value)
  • Hₐ: μ ≠ μ₀ (the population mean is different from that value)

Example: "Is the average test score different from 75?" You don't care if it's higher or lower, just different.

One-Tailed Tests (Upper-Tailed)

When you specifically want to test if a parameter is greater than a value:

  • H₀: μ ≤ μ₀ (or simply μ = μ₀)
  • Hₐ: μ > μ₀

Example: "Is the new drug more effective than the standard treatment?"

One-Tailed Tests (Lower-Tailed)

When you specifically want to test if a parameter is less than a value:

  • H₀: μ ≥ μ₀ (or simply μ = μ₀)
  • Hₐ: μ < μ₀

Example: "Does the new manufacturing process reduce defect rates?"

Key Principles

  • H₀ always includes equality (=, ≤, or ≥)
  • Hₐ never includes equality (<, >, or ≠)
  • The research claim usually goes in Hₐ
  • Choose one-tailed when direction matters; two-tailed when any change matters

Key Takeaway: The null hypothesis is the skeptical default we test against; the alternative is directional (one-tailed) or non-directional (two-tailed) depending on your research question.