Statistical hypothesis testing (AS Unit 2: Applied Mathematics A) — WJEC A-Level Mathematics
Statistical hypothesis testing (AS Unit 2: Applied Mathematics A) explained
Statistical hypothesis testing (AS Unit 2: Applied Mathematics A) covers 4 specification points, including Null hypothesis, alternative hypothesis,… and Conduct a statistical hypothesis test for the….
Read the full explanation
Open a point below for its explanation and revision guidance.
Understand and apply the language of statistical hypothesis testing, developed through a binomial model: null hypothesis, alternative hypothesis, significance level, test statistic, 1-tail test, 2-tail test, critical value, critical region, acceptance region, p-value.
Hypothesis testing provides a formal decision framework for evaluating claims about a population parameter. The null hypothesis H₀ states the default baseline (e.g. p = p₀), while the alternative hypothesis H₁ expresses what is suspected: one-tailed if directional (p < p₀ or p > p₀), two-tailed if non-directional (p ≠ p₀). The test statistic X is the observed count. The significance level α is the threshold probability for rejecting H₀ when true. The critical region is the set of test statistic values leading to rejection of H₀, bounded by critical values; its complement is the acceptance region. The p-value is the probability of observing a result at least as extreme as the test statistic under H₀.
Conduct a statistical hypothesis test for the proportion in the binomial distribution and interpret the results in context.
To execute a binomial hypothesis test: 1. State H₀: p = p₀ and H₁: p < p₀, p > p₀, or p ≠ p₀, defining p clearly in context. 2. State the significance level α (e.g. 5% or 1%). 3. Define the test statistic distribution under H₀: X ~ B(n, p₀). 4. Calculate the p-value: for an upper-tail test with observed x, compute P(X ≥ x); for a lower-tail test, compute P(X ≤ x); for a two-tailed test, double the tail probability or compare to α/2. 5. Compare the p-value to α (or check if x lies in the critical region). 6. Conclude in two parts: an explicit statistical decision (reject or do not reject H₀) followed by an interpretation in context using non-dogmatic phrasing.
Understand that a sample is being used to make an inference about the population and appreciate that the significance level is the probability of incorrectly rejecting the null hypothesis.
In statistical hypothesis testing, sample data provide empirical evidence to draw probabilistic inferences regarding an unknown population parameter. Because samples exhibit natural random variability, a test statistic can occasionally land in the critical region purely by chance even when the null hypothesis H₀ is entirely true. The significance level α specifies the maximum acceptable probability of committing this mistake—known as a Type I error. Candidates must recognise that selecting a smaller significance level (such as 1% instead of 5%) reduces the risk of incorrectly rejecting a true null hypothesis, but simultaneously increases the risk of failing to detect a real change when H₀ is false.
Interpret and calculate Type I and Type II errors, and know their practical meaning.
In decision theory, two distinct testing errors can occur: A Type I error occurs when the null hypothesis H₀ is rejected when H₀ is in fact true. Its probability is P(Type I error) = P(X ∈ Critical Region | H₀ is true), which equals the actual significance level. A Type II error occurs when H₀ is not rejected (accepted) when H₀ is in fact false (an alternative parameter value p₁ applies). Its probability is β = P(Type II error) = P(X ∉ Critical Region | H₁ is true). In practical applications—such as medical screening or manufacturing quality control—candidates interpret the real-world consequences of both errors and calculate their numerical probabilities.
Your focus
- Formulate null and alternative hypotheses using correct mathematical parameter notation.
- Distinguish between critical regions, acceptance regions, critical values, and significance levels.
- Explain how p-values and critical regions provide equivalent criteria for evaluating hypotheses.
Show all 12 objectives
- Conduct a complete hypothesis test for a binomial proportion using p-values or critical regions.
- Determine critical values for one-tailed and two-tailed binomial tests at specified significance levels.
- Formulate nuanced, non-dogmatic contextual conclusions based on statistical test outcomes.
- Explain the significance level as the probability of committing a Type I error.
- Calculate the actual significance level of a discrete test from its critical region.
- Evaluate the balance between Type I and Type II error probabilities when selecting a significance threshold.
- Define and distinguish between Type I (false positive) and Type II (false negative) statistical errors.
- Calculate the numerical probabilities of Type I and Type II errors for a specified binomial test.
- Assess the practical and financial consequences of Type I and Type II errors in real-world scenarios.
Statistical hypothesis testing (AS Unit 2: Applied Mathematics A) exam tips
Marking Points
- defining the null and alternative hypotheses using the population parameter p
- defining the critical region as the set of test statistic outcomes leading to the rejection of H₀
- explaining the role of the significance level and p-value in deciding whether to reject H₀
- stating H₀ and H₁ correctly with the population parameter p defined in context
- calculating the correct tail probability P(X ≥ x) or P(X ≤ x) under H₀
- comparing the tail probability to the significance level and stating whether to reject H₀
- writing a final contextual conclusion that avoids absolute or dogmatic phrasing
- explaining that the significance level is the probability of rejecting H₀ when H₀ is true
- noting that in discrete distributions, the actual significance level may be strictly less than the nominal α
- explaining the trade-off between reducing Type I error risk and increasing Type II error risk
- defining a Type I error as rejecting H₀ when H₀ is true, and a Type II error as accepting H₀ when H₀ is false
- calculating P(Type I error) as the probability of the critical region under H₀
- calculating P(Type II error) by evaluating the probability of the acceptance region under the alternative parameter p₁
- explaining the practical consequences of each error in the given contextual scenario
Examiner Tips
- 💡Always state hypotheses using the parameter symbol p (e.g. H₀: p = 0.3, H₁: p > 0.3), never sample values.
- 💡In a two-tailed test at significance level α, remember that each tail receives an error probability of α/2.
- 💡Always define p in words: 'where p is the probability that a component is defective'.
- 💡Write your conclusion in two sentences: first state whether to reject H₀, then write the contextual interpretation.
- 💡Remember that for discrete distributions, the actual significance level equals P(X in critical region | H₀ is true).
- 💡State clearly: 'The significance level is the probability of incorrectly rejecting a true null hypothesis.'
- 💡To calculate P(Type II error), evaluate the probability that the test statistic falls into the acceptance region using the NEW value of p.
- 💡Explain practical consequences clearly: a Type I error scraps good batches; a Type II error ships defective products.
Common Mistakes
- writing hypotheses using the sample proportion or sample statistic rather than the population parameter p
- confusing the significance level α with the calculated p-value or the critical value itself
- failing to double the calculated tail probability when conducting a two-tailed test using p-values
- writing dogmatic conclusions like 'proving the proportion has changed' rather than 'there is sufficient evidence to suggest...'
- believing that the significance level represents the probability that the null hypothesis is true
- confusing the nominal significance level (such as 5%) with the actual significance level determined by discrete critical regions
- calculating the Type II error probability under the null hypothesis instead of using the specified alternative parameter value
- mixing up which error corresponds to a false positive (Type I) versus a false negative (Type II)