Type I & Type II Errors - College Statistics

1. A factory's wastewater is tested to determine if the mean concentration of a pollutant exceeds the legally permissible limit of 100 parts per million (ppm). The regulatory agency sets up the hypotheses as H₀: μ ≤ 100 versus Hₐ: μ > 100. After analyzing water samples, the agency fails to reject the null hypothesis. If, in fact, the factory's mean pollutant level is 115 ppm, which error has been made and what is the consequence?
   Answer: A Type II error occurred, allowing the factory to continue polluting the environment at illegally high levels. A Type II error is failing to reject a false null hypothesis. Here, the agency failed to reject H₀ (μ ≤ 100). However, the reality is that the null hypothesis is false (μ = 115 ppm). This is a Type II error. The consequence is that the violation is not detected, and the factory continues to release excessive pollutants.
2. A 99% confidence interval for the difference between two population means (μ₁ - μ₂) is calculated to be (2.5, 8.3). If a hypothesis test were conducted for H₀: μ₁ - μ₂ = 0 versus Hₐ: μ₁ - μ₂ ≠ 0 using the same data, what would be the test's conclusion at the corresponding significance level, and what type of error could have been made?
   Answer: Reject H₀; a Type I error may have occurred. A 99% confidence interval corresponds to a hypothesis test at a significance level of α = 1 - 0.99 = 0.01. The null hypothesis H₀ states that the difference between the means is 0. Since the calculated confidence interval (2.5, 8.3) does not contain 0, the null hypothesis would be rejected. When the null hypothesis is rejected, it is possible that it was actually true, which means a Type I error may have occurred.
3. The graph shows the power curve for a hypothesis test of H₀: μ ≤ 50 versus Hₐ: μ > 50, conducted at the α = 0.05 significance level. According to the graph, if the true population mean is actually 55, what is the approximate probability of committing a Type II error?
   Answer: 0.25 The power curve plots the probability of rejecting H₀ for various true values of the parameter μ. The y-axis represents power (1 - β). First, locate the true mean μ = 55 on the x-axis. According to the graph, the corresponding y-value (power) is 0.75. A Type II error is failing to reject a false null hypothesis, and its probability is β. Since Power = 1 - β, we have β = 1 - Power = 1 - 0.75 = 0.25.
4. A clinical trial for a new cancer therapy is conducted with the null hypothesis that the therapy has no effect on patient survival rates. Which of the two error types is generally considered to have more severe human consequences, and how should this influence the choice of significance level α?
   Answer: Type II error is more severe; α should be set to a relatively high value (e.g., 0.10). When evaluating clinical trials, understanding the real-world consequences of statistical errors is crucial for making ethical decisions about significance levels. In this cancer therapy trial, a Type I error means rejecting a true null hypothesis—concluding the therapy works when it actually doesn't. A Type II error means failing to reject a false null hypothesis—concluding the therapy doesn't work when it actually does. From a human perspective, a Type II error is far more severe because it means withholding a potentially life-saving treatment from cancer patients who could benefit. To minimize Type II errors, you need to increase statistical power, which means being more willing to detect effects when they exist. This requires setting a higher significance level (like $α = 0.10$) rather than a very conservative one. A higher α makes it easier to reject the null hypothesis, reducing the chance of missing a beneficial treatment. Answer A incorrectly identifies Type I error as more severe—while approving an ineffective treatment has consequences, it's generally less catastrophic than missing a cure. Answer B correctly identifies Type II as more problematic but suggests the wrong α direction. Answer D makes the critical error of recommending a very low α, which would actually increase Type II error risk by making it harder to detect real treatment effects. Remember this pattern: in medical contexts where missing a beneficial treatment has severe consequences (like cancer, heart disease), researchers often accept higher Type I error risk to minimize Type II errors. The stakes determine your statistical strategy.
5. A company claims its new breakfast cereal contains 'at most 10 grams of sugar per serving'. A consumer advocacy group plans to test this claim, believing the sugar content is higher. They set up hypotheses H₀: μ ≤ 10 and Hₐ: μ > 10. For the consumer group, what would be the consequence of making a Type II error?
   Answer: The group fails to find evidence that the cereal has more than 10g of sugar when, in fact, it does. A Type II error is failing to reject a false null hypothesis. The null hypothesis (H₀) is that the mean sugar content is at most 10g (μ ≤ 10). If this is false, it means the true mean sugar content is greater than 10g (μ > 10). Failing to reject H₀ means the consumer group does not find sufficient evidence to challenge the company's claim. Therefore, a Type II error means the group fails to detect that the cereal's sugar content is higher than advertised.
6. In the American legal system, a defendant is presumed innocent until proven guilty. This corresponds to a hypothesis test where H₀: The defendant is innocent, and Hₐ: The defendant is guilty. The standard of "beyond a reasonable doubt" relates to the significance level. What does a Type II error represent in this context?
   Answer: Acquitting a guilty person. A Type II error is the failure to reject a false null hypothesis. In this analogy, the null hypothesis is innocence. If the null hypothesis is false, the defendant is actually guilty. Failing to reject H₀ means the jury does not find the defendant guilty, leading to an acquittal. Therefore, a Type II error is acquitting a person who is actually guilty.
7. A researcher performs a hypothesis test and decides to change the significance level α from 0.05 to 0.01. Assuming the sample size and the true mean remain the same, what is the definitive consequence for the probability of a Type II error (β) and the power of the test (1 - β)?
   Answer: β will increase and power will decrease. There is an inverse relationship between α and β. Decreasing α (from 0.05 to 0.01) makes the rejection criteria more stringent. This means it is harder to reject the null hypothesis. This reduces the probability of a Type I error (α) but increases the probability of failing to reject a false null hypothesis (β). Since power is defined as 1 - β, an increase in β leads to a decrease in power.
8. A study is designed to test H₀: μ = 50 versus Hₐ: μ > 50 with α = 0.05 and a sample of size n = 100. Researchers determine that if the true mean is actually μ = 52, the power of the test is 0.64. If, however, the true mean is actually μ = 54, what can be concluded about the power of the test?
   Answer: The power will be greater than 0.64. When you encounter questions about statistical power, remember that power is the probability of correctly rejecting a false null hypothesis. Power increases as the true parameter value moves further from the null hypothesis value in the direction of the alternative hypothesis. In this one-tailed test with $H_0: μ = 50$ versus $H_a: μ > 50$, you're given that when the true mean is $μ = 52$, the power is 0.64. If the true mean is actually $μ = 54$, this represents a larger effect size—the true parameter is now 4 units away from the null value instead of just 2 units. As the true mean moves further into the alternative hypothesis region (greater than 50), it becomes easier to detect this difference, increasing the probability of correctly rejecting the false null hypothesis. Choice A is incorrect because power increases, not decreases, with larger effect sizes. Choice B misunderstands the relationship—power depends on the effect size and other test parameters, not on p-values from individual samples. Choice C reflects a common misconception that power is fixed when α and n are constant. While these factors do influence power, the effect size (difference between true and hypothesized values) is equally important in determining power. The answer is D: the power will be greater than 0.64. Study tip: Remember that power always increases as the true parameter moves further from the null hypothesis value in the direction specified by the alternative hypothesis. Visualize the power curve—it's steepest near the null value and approaches 1 as effect size increases.
9. A researcher plans a study to test if a new teaching method improves test scores from the current mean of 75. The researcher is certain the new method will not worsen scores and decides to use a one-tailed test (Hₐ: μ > 75) instead of a two-tailed test (Hₐ: μ ≠ 75), keeping α and the sample size the same. How does this choice affect the power of the test to detect a true mean score of, for example, 80?
   Answer: The one-tailed test will have more power because the entire rejection region is placed in the upper tail. When comparing one-tailed and two-tailed hypothesis tests, the key insight is understanding how the rejection region affects your ability to detect true effects. Statistical power is the probability of correctly rejecting a false null hypothesis, and it's directly influenced by how you allocate your critical region. In a two-tailed test with $H_a: \mu \neq 75$, you split your significance level $\alpha$ between both tails ($\alpha/2$ in each tail). For a one-tailed test with $H_a: \mu > 75$, you place the entire $\alpha$ in the upper tail. This means the critical value for the one-tailed test is less extreme than for the two-tailed test, making it easier to reject the null hypothesis when the true mean is indeed greater than 75. Since the researcher is testing whether the true mean is 80 (which is greater than 75), the one-tailed test concentrates all its "rejection power" exactly where the effect is expected to occur. This increases the probability of detecting the difference. Choice A is wrong because not considering $\mu < 75$ doesn't reduce power when testing for $\mu > 75$—it actually helps by focusing the test. Choice B is incorrect because power comparisons between these tests don't require knowing the population standard deviation; the relationship holds regardless. Choice C misses the crucial point that while the total $\alpha$ is the same, its distribution across tails differs, directly affecting power. Study tip: Remember that one-tailed tests always have more power than two-tailed tests when testing in the correct direction, because you're concentrating your entire $\alpha$ where you expect to find the effect.
10. A pharmaceutical company develops a new drug to treat a rare, life-threatening disease for which the current treatment has severe side effects. The company conducts a clinical trial to test the new drug's efficacy. The null hypothesis is that the new drug is no more effective than the current treatment. If a Type I error is committed during this trial, what is the most direct and significant consequence?
    Answer: The new drug, which is no more effective than the current treatment, will be approved and marketed, potentially exposing patients to new risks without benefit. A Type I error is the rejection of a true null hypothesis. In this context, the null hypothesis (H₀) is that the new drug is no more effective. Rejecting H₀ means concluding the new drug is more effective. If this is an error, it means the new drug is actually not an improvement. Therefore, the consequence is marketing a new drug that is no better than the old one, which could expose patients to unknown side effects for no additional therapeutic benefit.
11. An experiment with a fixed sample size yields a p-value of 0.055 for a test of H₀: μ = 30 versus Hₐ: μ > 30. The researcher had set the significance level α = 0.05. The researcher is concerned that a small but important effect might actually exist. Which of the following is the most statistically sound explanation for the failure to find a significant result?
    Answer: The study may have been underpowered, possibly due to an insufficient sample size to detect the small effect. A p-value (0.055) that is very close to the significance level (0.05) suggests that there might be a real effect, but the evidence was not quite strong enough to be declared statistically significant. A common reason for this is low statistical power. The study may not have had a large enough sample size to reliably detect the true effect, leading to a potential Type II error. This is a more plausible explanation than the others.
12. All other factors being held constant, which of the following scenarios will result in the hypothesis test with the greatest statistical power?
    Answer: A large sample size, a large effect size, and α = 0.05. Statistical power measures your ability to correctly reject a false null hypothesis—essentially, how likely you are to detect a real effect when it exists. Three key factors determine power: sample size, effect size, and significance level (α). Larger sample sizes increase power because they reduce sampling variability, making it easier to distinguish between the null hypothesis and alternative hypothesis. When you have more data points, your estimates become more precise and reliable. Larger effect sizes also increase power because bigger differences are easier to detect. If the true effect is substantial, your test is more likely to pick it up compared to a tiny, subtle effect. Finally, higher α levels (like 0.05 vs. 0.01) increase power because you're using a less stringent threshold for rejection. While this increases Type I error risk, it makes it easier to detect true effects. Answer C combines all three power-boosting factors: large sample size, large effect size, and α = 0.05. This creates optimal conditions for detecting a true effect. Answer A has the worst combination—small sample size and small effect size severely limit power, while α = 0.01 makes detection even harder. Answer B improves with a large effect size and higher α, but the small sample size still constrains power significantly. Answer D has a large sample size but is hampered by the small effect size and very stringent α = 0.01 threshold. Remember: when maximizing statistical power, you want "big, big, big"—big sample size, big effect size, and bigger α (within reasonable limits).
13. The statistical power of a hypothesis test is calculated to be 0.80 for a specific alternative hypothesis. Which statement is the correct interpretation of this value?
    Answer: If this specific alternative hypothesis is true, there is an 80% probability of correctly rejecting the null hypothesis. Statistical power is the probability of correctly rejecting a false null hypothesis. It is a conditional probability: P(Reject H₀ | H₀ is false). The value of power depends on the specific value of the parameter under the alternative hypothesis. So, a power of 0.80 means that if the parameter has that specific alternative value, the experiment has an 80% chance of producing a statistically significant result.
14. A quality control engineer at a bottling plant tests H₀: μ = 12 ounces versus Hₐ: μ ≠ 12 ounces for the mean volume of filled bottles. A Type I error occurs if the engineer concludes that the mean volume is...
    Answer: not equal to 12 ounces when, in fact, it is 12 ounces. When you encounter hypothesis testing questions, focus on the precise definitions of Type I and Type II errors. A Type I error occurs when you reject a true null hypothesis, while a Type II error occurs when you fail to reject a false null hypothesis. In this bottling plant scenario, the null hypothesis states that the mean volume equals 12 ounces (H₀: μ = 12). The alternative hypothesis claims it doesn't equal 12 ounces (Hₐ: μ ≠ 12). A Type I error happens when you reject H₀ when it's actually true. This means concluding the mean volume is not equal to 12 ounces when it actually is 12 ounces. Answer A correctly describes this scenario: concluding the mean is not equal to 12 ounces when it actually is 12 ounces. This is exactly what happens in a Type I error. Answer B describes a Type II error instead—failing to reject a false null hypothesis. Here, you'd conclude the mean equals 12 ounces when it actually doesn't. Answer C represents a correct decision, not an error. If the mean truly isn't 12 ounces and you conclude it's not 12 ounces, you've made the right call. Answer D also represents a correct decision. If the mean truly is 12 ounces and you conclude it is 12 ounces, you've correctly failed to reject a true null hypothesis. Study tip: Remember "Type I = reject when true." Create a mental connection between Type I errors and false alarms—you're detecting a problem that doesn't actually exist, just like a smoke detector going off when there's no fire.
15. A research team is conducting a hypothesis test where the consequences of a Type I error are judged to be far more severe than the consequences of a Type II error (e.g., approving a harmful new drug). Which of the following represents the most appropriate strategy for designing their test?
    Answer: Set a very low significance level (e.g., α = 0.001) to minimize the chance of incorrectly rejecting the null hypothesis. The probability of a Type I error is denoted by the significance level, α. If a Type I error is very costly, the researcher should make its probability as low as possible. Choosing a very small α (like 0.001) directly reduces the risk of committing this type of error. While this will increase the probability of a Type II error (β), the question states that this is the less severe error.
16. A significance level of α = 0.05 is used in a hypothesis test. Which statement provides the most accurate definition of what this α value represents?
    Answer: The probability of rejecting the null hypothesis, assuming the null hypothesis is true, is 5%. The significance level α is the conditional probability of making a Type I error. A Type I error is rejecting the null hypothesis (H₀) when it is actually true. Therefore, α = P(Reject H₀ | H₀ is true). Choice B is a common misconception; it describes P(H₀ is true | Reject H₀), which is not α. Choice C describes the p-value, not α.
17. A political campaign wants to test if a new ad series has increased a candidate's approval rating above 40%. The null hypothesis is H₀: p ≤ 0.40. The campaign is concerned about failing to detect a genuine increase in approval. To address this, they wish to reduce the probability of a Type II error without changing the pre-specified significance level (α). Which action would achieve this?
    Answer: Increasing the number of voters surveyed for the poll. The probability of a Type II error (β) is reduced by increasing the power of the test. The three main factors that increase power are increasing the significance level (α), increasing the effect size, or increasing the sample size (n). Since the campaign cannot change α or the true effect size, the only action they can take is to increase the sample size. A larger sample provides more information and makes it easier to detect a true effect.
18. In a hypothesis test of H₀: μ = 10 against Hₐ: μ ≠ 10, a p-value of 0.12 is obtained from a sample. A researcher using a significance level of α = 0.05 fails to reject H₀. Which of the following statements correctly describes the situation?
    Answer: The researcher has failed to find sufficient evidence that the true mean is not 10, and it is possible a Type II error was committed. Since the p-value (0.12) is greater than α (0.05), the correct decision is to fail to reject H₀. When we fail to reject H₀, we never prove H₀ is true; we only conclude there is insufficient evidence against it. The only type of error that can be made in this situation is a Type II error, which occurs if we fail to reject H₀ when it is actually false.
19. The diagram below illustrates the sampling distributions under a null hypothesis H₀ and a specific alternative hypothesis Hₐ. The vertical line represents the critical value for the decision rule, and the shaded area labeled 'α' represents the probability of a Type I error. Which area in the diagram represents β, the probability of a Type II error?
    Answer: The area under the Hₐ curve to the left of the critical value. A Type II error (β) is the probability of failing to reject the null hypothesis when the alternative hypothesis is true. The decision rule is to 'fail to reject H₀' for any sample mean to the left of the critical value. The probability of this happening, given that Hₐ is true, is represented by the area under the Hₐ distribution curve that falls within the 'fail to reject' region, which is the area to the left of the critical value.
20. For a hypothesis test with a fixed sample size and a fixed, non-zero effect size, consider a graph where the probability of a Type II error (β) is on the y-axis and the probability of a Type I error (α) is on the x-axis. As α is allowed to vary from 0 to 1, which of the following best describes the shape of the resulting curve?
    Answer: A strictly decreasing curve, showing the trade-off between the two error types. When you encounter questions about Type I and Type II errors, focus on the fundamental relationship between these error types. Type I error (α) is rejecting a true null hypothesis, while Type II error (β) is failing to reject a false null hypothesis. The key insight is that α and β have an inverse relationship when sample size and effect size are fixed. As you make your test more stringent by decreasing α (requiring stronger evidence to reject the null), you simultaneously make it harder to detect a true effect, thus increasing β. Conversely, as you increase α (making it easier to reject the null), you become more likely to detect a true effect when one exists, decreasing β. This creates a strictly decreasing curve where β decreases as α increases - exactly what option A describes. The relationship isn't linear, but it's monotonically decreasing. Option B incorrectly suggests both error types increase together, which contradicts their inverse relationship. Option C proposes a U-shaped curve implying β could be minimized at some moderate α value, but this isn't true - β continues decreasing as α increases (though with diminishing returns). Option D suggests independence between the errors, which ignores the fundamental trade-off that defines hypothesis testing. Remember this core principle: with fixed sample size and effect size, you cannot simultaneously minimize both error types. This trade-off is why choosing an appropriate significance level (typically 0.05) involves balancing the costs of each error type for your specific situation.

Start this quiz