Acceptance Region

A key concept in statistical inference defining the range where a test statistic is likely to fall if the null hypothesis is true
On this page

In one sentence

The acceptance region is the set of test statistic values for which you do not reject the null hypothesis at a chosen significance level.

Definition

In a hypothesis test, you choose a significance level \(\alpha\) (Type I error rate). The rejection region is constructed so that:

\[ \mathbb{P}(\text{reject } H_0 \mid H_0 \text{ true}) = \alpha \]

The acceptance region is the complement of the rejection region.

One-sided vs two-sided tests

  • One-sided test: rejection region is in one tail (e.g., large values only).
  • Two-sided test: rejection region is split across both tails (extreme high or low values).

A concrete example (critical values)

For a two-sided z-test, you reject \(H_0\) for “too large” or “too small” values of the test statistic \(Z\). The acceptance region is:

\[ -z_{1-\alpha/2} \le Z \le z_{1-\alpha/2} \]

Relationship to p-values and confidence intervals

  • If \(p \le \alpha\), the test statistic falls in the rejection region and you reject \(H_0\).
  • Many tests have an equivalent confidence interval view: reject \(H_0: \theta = \theta_0\) at level \(\alpha\) if \(\theta_0\) lies outside the \((1-\alpha)\) confidence interval.

Confidence-interval equivalence (quick example)

For testing a population mean with known variance using a z-test, the \((1-\alpha)\) confidence interval is:

\[ \bar X \pm z_{1-\alpha/2}\,\frac{\sigma}{\sqrt{n}} \]

You reject $H_0: \mu = \mu_0$ at level $\alpha$ exactly when $\mu_0$ lies outside that interval—this is the same decision rule expressed as “rejection region” vs “confidence interval.”

Visual map

    flowchart TD
	  A["Choose alpha (significance level)"] --> B["Compute test statistic"]
	  B --> C{"Statistic in rejection region?"}
	  C -- "Yes" --> D["Reject H0"]
	  C -- "No" --> E["Fail to reject H0<br/>(in acceptance region)"]
	  D --> F["Report p-value and context"]
	  E --> F
  • Rejection Region: The set of all values of the test statistic for which the null hypothesis is rejected.
  • Null Hypothesis: A general statement or default position that there is no relationship between two measured phenomena.
  • Test Statistic: A standardized value calculated from sample data during a hypothesis test.

Quiz

### What is an acceptance region in statistical inference? - [x] The range of values of the test statistic where the null hypothesis cannot be rejected. - [ ] The range of values of the test statistic where the null hypothesis is always rejected. - [ ] A measure of the probability against the null hypothesis. - [ ] A set value that defines the null hypothesis. > **Explanation:** The acceptance region is where the test statistic falls to support the null hypothesis. ### Which term is the complement of the acceptance region? - [ ] P-value - [x] Rejection region - [ ] Critical region - [ ] Confidence interval > **Explanation:** The complement of the acceptance region is the rejection region. ### True or False: The null hypothesis is always rejected if the test statistic falls within the acceptance region. - [ ] True - [x] False > **Explanation:** The null hypothesis is NOT rejected if the test statistic falls within the acceptance region. ### Which factor directly affects the size of the acceptance region? - [ ] Sample mean - [x] Significance level (α) - [ ] Population variance - [ ] Test statistic magnitude > **Explanation:** The size of the acceptance region is influenced by the chosen significance level ($\alpha$). ### In a two-sided z-test at level $\alpha$, what is the acceptance region for $Z$? - [x] $-z_{1-\alpha/2} \le Z \le z_{1-\alpha/2}$ - [ ] $Z \ge z_{1-\alpha/2}$ - [ ] $Z \le -z_{1-\alpha/2}$ - [ ] $-z_{\alpha/2} \le Z \le z_{\alpha/2}$ > **Explanation:** Two-sided tests split $\alpha$ across both tails, leaving the middle $1-\alpha$ probability mass as the acceptance region. ### If a test has p-value $p$ and you use significance level $\alpha$, when do you reject $H_0$? - [x] When $p \le \alpha$ - [ ] When $p \ge \alpha$ - [ ] Only when $p = 0$ - [ ] Only when $p = 1$ > **Explanation:** By definition, the p-value is the smallest $\alpha$ at which you would reject $H_0$. ### True or False: “Fail to reject $H_0$” is the same as “$H_0$ is true.” - [ ] True - [x] False > **Explanation:** Failing to reject means the data are not sufficiently inconsistent with $H_0$ at the chosen $\alpha$; it does not prove $H_0$. ### How does increasing $\alpha$ (e.g., from 0.05 to 0.10) affect the acceptance region? - [x] It shrinks the acceptance region and expands the rejection region. - [ ] It expands the acceptance region and shrinks the rejection region. - [ ] It has no effect. - [ ] It changes the test statistic definition > **Explanation:** A larger $\alpha$ allows more Type I error, so the rejection region must get larger. ### Which statement best links the acceptance region to a confidence interval? - [x] Reject $H_0: \theta=\theta_0$ at level $\alpha$ if $\theta_0$ is outside the $(1-\alpha)$ confidence interval. - [ ] Reject $H_0$ if the confidence interval is wide. - [ ] Reject $H_0$ whenever $\alpha$ is small. - [ ] Confidence intervals are unrelated to hypothesis tests. > **Explanation:** Many standard tests have an equivalent CI formulation; it’s the same decision rule expressed differently. ### What happens when the test statistic falls outside the acceptance region? - [x] The null hypothesis is rejected. - [ ] The hypothesis remains untested. - [ ] The test must be discarded. - [ ] The acceptance region is recalculated. > **Explanation:** If the test statistic falls outside the acceptance region, the null hypothesis is typically rejected.
Accepting House
Acceptance