Study guide: One-Sample Hypothesis Testing

Course Outline

  1. Statistical Hypotheses and Test Outcomes
  2. Errors and Significance Level
  3. Test Decisions and Procedures
  4. One-Sample Z-Test for a Mean
  5. P-Value Decisions and Interpretation
  6. T-Tests for a Population Mean
  7. Z-Tests for a Population Proportion
  8. Confidence Intervals and Hypothesis Tests
  9. Parameters for Two Independent Samples
  10. Critical Values and Rejection Regions
  11. Large-Sample Mean Inference
  12. Small-Sample Mean Inference
  13. Large-Sample Proportion Inference
  14. Chi-Squared Test of Independence
  15. Standardized Residual Analysis

1. Statistical Hypotheses and Test Outcomes

Key Concepts & Definitions

  • Statistical hypothesis: A claim about the numerical value of a population parameter.
  • Null hypothesis: The null hypothesis, denoted Hβ‚€, is the hypothesis tested and usually represents the status quo or a claim about a population parameter, commonly that the parameter equals a hypothesized value.
  • Alternative hypothesis: The alternative hypothesis, denoted Hₐ or H₁, is the opposite of Hβ‚€ and is accepted only when sample data provide convincing evidence for it.

β˜… Must-know

πŸ“Œ Rejecting Hβ‚€ provides evidence supporting Hₐ, whereas failing to reject Hβ‚€ means there is insufficient evidence to support Hₐ and does not establish that Hβ‚€ is true.

Further detail

  • The three forms are:
    • Two-tailed: parameter β‰  hypothesized value
    • Lower-tailed: parameter < hypothesized value
    • Upper-tailed: parameter > hypothesized value

Memory Hook

Rejecting Hβ‚€ supports Hₐ; failing to reject Hβ‚€ does not prove Hβ‚€.

2. Errors and Significance Level

Key Concepts & Definitions

  • Type I error: A Type I error occurs when a true null hypothesis is rejected, and its probability is Ξ±, the test's significance level set in advance by the researcher.
  • Type II error: A Type II error occurs when a false null hypothesis is not rejected, and its probability is Ξ².

Essential Points

  • For a disease test with Hβ‚€ stating no disease, a Type I error is a false positive and a Type II error is a false negative.

3. Test Decisions and Procedures

Key Concepts & Definitions

  • P-value: The conditional probability, assuming Hβ‚€ is true, of obtaining a test statistic equal to or more extreme than the observed value; it is also called the observed level of significance.
  • Test statistic: A sample statistic computed from sample information and used to decide between the null and alternative hypotheses.

β˜… Must-know

πŸ“Œ A critical-value method decides by checking whether the test statistic falls in a rejection region, whereas a p-value method compares the p-value with Ξ±.

πŸ“Œ When the p-value is greater than Ξ±, do not reject Hβ‚€; when the p-value is less than Ξ±, reject Hβ‚€.

  • A hypothesis test proceeds by stating Hβ‚€ and Hₐ, checking assumptions, finding the test statistic, finding critical values or a p-value, and concluding in context.

Further detail

  • For a two-tailed test, the p-value is twice the corresponding one-tailed p-value.

  • The sampling distribution under Hβ‚€ is used to judge whether an observed sample statistic is sufficiently far from the hypothesized value to reject Hβ‚€.

Memory Hook

State, check, calculate, compare, conclude.

4. One-Sample Z-Test for a Mean

β˜… Must-know

πŸ“ Formula - For a one-sample z-test of a population mean with known population standard deviation, the test statistic is ZSTAT=XΛ‰βˆ’ΞΌ0Οƒ/nZ_{STAT}=\frac{\bar{X}-\mu_0}{\sigma/\sqrt{n}}.

  • For the bolt example, the test statistic is ZSTAT = βˆ’2.0 and the two-tailed critical values at Ξ± = 0.05 are βˆ’1.96 and 1.96, so Hβ‚€ is rejected because βˆ’2.0 lies in the rejection region.

Further detail

πŸ“Œ If the population standard deviation Οƒ is unknown in the stated z-test, it is estimated with the sample standard deviation s.

  • For z-tests, critical values are obtained from a z-table or the last row of a t-table, using the significance level Ξ±.

  • The listed z-test critical values are: at Ξ± = 10%, lower-tailed βˆ’1.28 and upper-tailed 1.28; at Ξ± = 5%, lower-tailed βˆ’1.645 and upper-tailed 1.645, or two-tailed Β±1.96; at Ξ± = 1%, lower-tailed βˆ’2.33 and upper-tailed 2.33, or two-tailed Β±2.58.

πŸ“Œ The conditions for inference with a z-test for a mean are the same as those for a z-interval for the mean.

  • In the bolt example, a random sample of 100 bolts has mean diameter 29.84 mm, the claimed population mean is 30 mm, and the assumed population standard deviation is 0.8 mm.

  • For the bolt example, the hypotheses are Hβ‚€: ΞΌ = 30 mm and Hₐ: ΞΌ β‰  30 mm, with Ξ± = 0.05 because no direction is specified.

5. P-Value Decisions and Interpretation

β˜… Must-know

πŸ“Œ When the p-value is lower than the significance level Ξ±, reject the null hypothesis; when it is higher, fail to reject the null hypothesis.

πŸ“Œ A high p-value means the sample is not sufficiently inconsistent with the null hypothesis, but it does not prove that the null hypothesis is true.

Further detail

  • For a two-tailed z-test with test statistic βˆ’2.0, the p-value is P(Z < βˆ’2.0) + P(Z > 2.0) = 0.0228 + 0.0228 = 0.0456.

  • For the bolt-diameter test of Hβ‚€: ΞΌ = 30 mm, a p-value of 0.0456 is below Ξ± = 0.05, so the null hypothesis is rejected.

Memory Hook

A low p-value challenges Hβ‚€; a high p-value does not.

6. T-Tests for a Population Mean

β˜… Must-know

πŸ“ Formula - When the population standard deviation is unknown and the sample is small, the mean-test statistic is tstat=xΛ‰βˆ’ΞΌ0s/nt_{\text{stat}}=\frac{\bar{x}-\mu_0}{s/\sqrt{n}}.

πŸ“Œ For a one-sample t-test, critical values or p-values are determined from the t-distribution using degrees of freedom df=nβˆ’1df=n-1 and significance level Ξ±.

Further detail

  • In the hotel-room example, a sample of 25 hotels had a mean nightly cost of $172.50 and a standard deviation of $15.40; testing Hβ‚€: ΞΌ = $168 against Hₐ: ΞΌ > $168 gave tstat=1.46t_{\text{stat}}=1.46 with 24 degrees of freedom.

  • For the hotel-room test at Ξ± = 0.05, the critical t-value with 24 degrees of freedom is 1.711, so the statistic 1.46 does not fall in the rejection region.

  • For the hotel-room test, the p-value is between 5% and 10%; statistical software gives 7.85%, which exceeds Ξ± = 0.05, so the null hypothesis is not rejected.

  • An incomplete t-table may only bound a t-test p-value between two significance levels, while software can provide an exact p-value.

7. Z-Tests for a Population Proportion

β˜… Must-know

πŸ“ Formula - For a one-sample proportion z-test, the test statistic is Zstat=p^βˆ’p0p0(1βˆ’p0)/nZ_{\text{stat}}=\frac{\hat{p}-p_0}{\sqrt{p_0(1-p_0)/n}}.

πŸ“Œ The normal approximation for a proportion test is used when both np0β‰₯15np_0\geq 15 and n(1βˆ’p0)β‰₯15n(1-p_0)\geq 15; if both quantities are below 15, the method is not discussed.

Further detail

  • In the mailing-list example, 25 of 500 clients responded, giving a sample proportion of 5%; testing Hβ‚€: p = 8% against Hₐ: p < 8% gives Zstat=βˆ’2.47Z_{\text{stat}}=-2.47.

  • For the mailing-list test, the p-value is 0.0068, which is below Ξ± = 0.05, so the null hypothesis is rejected in favor of a response rate below 8%.

8. Confidence Intervals and Hypothesis Tests

Essential Points

πŸ“Œ After rejecting a null hypothesis, a confidence interval based on the sample statistic can help indicate plausible values of the population parameter; using the same significance level, the null-hypothesis value is expected to lie outside the interval.

9. Parameters for Two Independent Samples

Essential Points

πŸ“Œ For comparing two independent samples, the target parameter is ΞΌ1βˆ’ΞΌ2\mu_1-\mu_2 for a difference in means with quantitative data, and p1βˆ’p2p_1-p_2 for a difference in proportions with qualitative data.

Memory Hook

Means compare quantitative data; proportions compare qualitative data.

10. Critical Values and Rejection Regions

Essential Points

  • For a z-test at significance levels 10%, 5%, and 1%, the lower-tailed critical values are βˆ’1.28, βˆ’1.645, and βˆ’2.33, and the upper-tailed values are 1.28, 1.645, and 2.33.

  • For a two-tailed z-test, the critical values at significance levels 10%, 5%, and 1% are Β±1.645, Β±1.96, and Β±2.58, respectively; the same critical values are used for confidence intervals.

11. Large-Sample Mean Inference

β˜… Must-know

πŸ“Œ Large-sample inference about μ₁ βˆ’ ΞΌβ‚‚ requires independently and randomly selected samples from the two target populations, with n₁ β‰₯ 30 and nβ‚‚ β‰₯ 30.

πŸ“ Formula - The large-sample confidence interval for μ₁ βˆ’ ΞΌβ‚‚ is (xΛ‰1βˆ’xΛ‰2)Β±zΞ±/2s12n1+s22n2\left(\bar{x}_1-\bar{x}_2\right)\pm z_{\alpha/2}\sqrt{\frac{s_1^2}{n_1}+\frac{s_2^2}{n_2}}.

πŸ“ Formula - For a large-sample test of μ₁ βˆ’ ΞΌβ‚‚, the test statistic is z=(xΛ‰1βˆ’xΛ‰2)βˆ’(ΞΌ1βˆ’ΞΌ2)0s12n1+s22n2z=\frac{(\bar{x}_1-\bar{x}_2)-(\mu_1-\mu_2)_0}{\sqrt{\frac{s_1^2}{n_1}+\frac{s_2^2}{n_2}}}, using sample standard deviations to estimate the population standard deviations.

Further detail

  • For each pair of independent samples, calculate the difference between the two sample means; the sampling distribution of these differences is centered on the population mean difference μ₁ βˆ’ ΞΌβ‚‚.

  • For the NYSE and NASDAQ samples (n₁ = 121, nβ‚‚ = 125; means 3.27 and 2.53; standard deviations 1.30 and 1.16), the 95% confidence interval for the mean dividend-yield difference, NYSE minus NASDAQ, is 0.43 to 1.05.

πŸ“Œ A confidence interval for μ₁ βˆ’ ΞΌβ‚‚ that is entirely positive indicates a higher population mean for Population 1, whereas an interval containing zero does not establish a significant difference between the population means.

  • For the one-tailed test comparing NASDAQ minus NYSE mean dividend yield at Ξ± = 0.05, the test statistic is z = βˆ’4.71 and the critical value is βˆ’1.645, so the null hypothesis is rejected and the NASDAQ population mean yield is significantly lower.

12. Small-Sample Mean Inference

β˜… Must-know

πŸ“Œ Small-sample inference about μ₁ βˆ’ ΞΌβ‚‚ requires independent random samples, approximately normal distributions in both populations, and equal population variances.

πŸ“ Formula - The small-sample confidence interval for μ₁ βˆ’ ΞΌβ‚‚ is (xΛ‰1βˆ’xΛ‰2)Β±tΞ±/2, n1+n2βˆ’2sp1n1+1n2\left(\bar{x}_1-\bar{x}_2\right)\pm t_{\alpha/2,\,n_1+n_2-2}s_p\sqrt{\frac{1}{n_1}+\frac{1}{n_2}}, with n₁ + nβ‚‚ βˆ’ 2 degrees of freedom.

πŸ“ Formula - The small-sample test statistic for μ₁ βˆ’ ΞΌβ‚‚ is t=(xΛ‰1βˆ’xΛ‰2)βˆ’(ΞΌ1βˆ’ΞΌ2)0sp1n1+1n2t=\frac{(\bar{x}_1-\bar{x}_2)-(\mu_1-\mu_2)_0}{s_p\sqrt{\frac{1}{n_1}+\frac{1}{n_2}}}, with n₁ + nβ‚‚ βˆ’ 2 degrees of freedom.

Further detail

πŸ“ Formula - The pooled variance for two independent samples with equal population variances is sp2=(n1βˆ’1)s12+(n2βˆ’1)s22n1+n2βˆ’2s_p^2=\frac{(n_1-1)s_1^2+(n_2-1)s_2^2}{n_1+n_2-2}.

  • For the NYSE and NASDAQ samples (n₁ = 11, nβ‚‚ = 15; means 3.27 and 2.53; standard deviations 1.30 and 1.16), assuming normal populations and equal variances, the 95% confidence interval for μ₁ βˆ’ ΞΌβ‚‚ is βˆ’0.26 to 1.74, with 24 degrees of freedom.

πŸ“Œ A confidence interval for μ₁ βˆ’ ΞΌβ‚‚ that contains zero does not establish a significant difference between the population means because the difference could be positive, negative, or zero.

  • A t-table that lacks exact p-values can only bound the p-value between significance levels, so the critical-value method is preferred unless software provides an exact p-value.

13. Large-Sample Proportion Inference

β˜… Must-know

πŸ“Œ Large-sample inference about p₁ βˆ’ pβ‚‚ requires independent random samples and, in each sample, at least 15 expected successes and 15 expected failures, checked using n₁p̂₁, n₁q̂₁, nβ‚‚pΜ‚β‚‚, and nβ‚‚qΜ‚β‚‚.

πŸ“ Formula - The large-sample confidence interval for p₁ βˆ’ pβ‚‚ has the form (p^1βˆ’p^2)Β±zΞ±/2 SE\left(\hat{p}_1-\hat{p}_2\right)\pm z_{\alpha/2}\,SE.

  • Because the confidence interval from 0.029 to 0.391 is entirely positive, the population proportion rating Method 1 fair is significantly higher than the proportion for Method 2.

πŸ“ Formula - For testing equality of two population proportions, the pooled estimate is p^=x1+x2n1+n2\hat p=\frac{x_1+x_2}{n_1+n_2} and the test statistic is z=p^1βˆ’p^2p^(1βˆ’p^)(1/n1+1/n2)z=\frac{\hat p_1-\hat p_2}{\sqrt{\hat p(1-\hat p)(1/n_1+1/n_2)}}.

πŸ“Œ For a two-proportion test, the null hypothesis states that the population proportions are equal, while the alternative may state that they differ, that population 1 is smaller, or that population 1 is larger.

  • At significance level 0.01, the employee fairness test gives z = 2.90, exceeding the one-sided critical value 2.33, so the null hypothesis is rejected and Method 1 has the higher population fair-rating proportion.

Further detail

  • For each pair of independent samples, calculate the difference between the sample proportions; the sampling distribution of these differences is centered on the population proportion difference p₁ βˆ’ pβ‚‚.

  • For 63 of 78 employees rating Method 1 fair and 49 of 82 rating Method 2 fair, the 99% confidence interval for the difference in population proportions is 0.029 to 0.391.

  • In the employee fairness test, the sample proportions are 63/78 = 0.808 and 49/82 = 0.598, and the pooled estimate is 112/160 = 0.70.

14. Chi-Squared Test of Independence

Key Concepts & Definitions

  • Chi-squared test of independence: A test assessing whether two categorical variables are independent in the population from which a sample is drawn.

β˜… Must-know

πŸ“ Formula - The chi-squared statistic is Ο‡2=βˆ‘iβˆ‘j(Oijβˆ’Eij)2Eij\chi^2=\sum_i\sum_j\frac{(O_{ij}-E_{ij})^2}{E_{ij}}, where the expected count in cell (i,j)(i,j) is Eij=RiCjnE_{ij}=\frac{R_iC_j}{n}, with row total RiR_i, column total CjC_j, and sample size nn.

πŸ“Œ For a test of independence, the null hypothesis states that the variables are independent in the population, while the alternative states that they are not independent.

πŸ“Œ The chi-squared independence test requires counts from a random sample and sufficiently large expected counts, with at least 80% of expected counts at least 5.

Further detail

πŸ“ Formula - For a contingency table with R rows and C columns, the degrees of freedom are df=(Rβˆ’1)(Cβˆ’1)df=(R-1)(C-1).

  • In the tattoo and hepatitis study, the chi-squared statistic is 57.91 with 2 degrees of freedom; at significance level 0.05 the critical value is 5.99 and the p-value is below 0.0001, so independence is rejected.

15. Standardized Residual Analysis

β˜… Must-know

πŸ“ Formula - A cell's standardized residual is Oβˆ’EE\frac{O-E}{\sqrt{E}}, where OO is the observed count and EE is the expected count.

  • A positive standardized residual means more observations were observed than expected under independence, while a negative residual means fewer were observed than expected.

Further detail

πŸ“Œ Standardized residuals greater than +2 or less than -2 are considered extreme.

  • In the tattoo and hepatitis example, the parlor/Hepatitis C cell has a standardized residual of +6.628, indicating more cases than expected, while the no-tattoo/Hepatitis C cell has a residual of -2.661, indicating fewer cases than expected.

Synthesis Tables

Two-Sample Target Parameters

ComparisonParameterData type
Difference in meansΞΌ1βˆ’ΞΌ2\mu_1-\mu_2Quantitative
Difference in proportionsp1βˆ’p2p_1-p_2Qualitative

Test yourself

Take a 38-question practice quiz on One-Sample Hypothesis Testing, with every answer explained.

1. Which statement qualifies as a statistical hypothesis?

2. A researcher tests whether the population mean waiting time equals 12 minutes. Which statement is an appropriate null hypothesis?

Take the quiz β†’

Study with flashcards

Memorize the key concepts of One-Sample Hypothesis Testing with 87 flashcards.

What does a statistical hypothesis claim about a population parameter?

Its numerical value.

What does the null hypothesis, denoted Hβ‚€, usually represent?

The status quo or a claim about a population parameter.

When is the alternative hypothesis, denoted Hₐ or H₁, accepted?

When sample data provide convincing evidence for it.

See the flashcards β†’

Related classes

Make your own study guides

Upload your class notes and Revizly writes a study guide, a quiz and flashcards in about a minute.

Study guide maker