Statistical Hypothesis Testing — OCR A-Level Mathematics

Statistical hypothesis testing is a core component of OCR A-Level Mathematics (H240), typically studied in Year 13. It provides a formal framework for making decisions about population parameters based on sample data. The process involves stating a null hypothesis (H₀) and an alternative hypothesis (H₁), selecting a significance level (usually 5% or 1%), calculating a test statistic, and comparing it to a critical value or using a p-value to decide whether to reject H₀. This topic builds on probability distributions, particularly the binomial and normal distributions, and is essential for understanding how conclusions are drawn in real-world contexts such as medicine, psychology, and quality control.

Hypothesis testing is not just a mathematical exercise; it teaches critical thinking about evidence and uncertainty. Students learn to quantify the strength of evidence against a claim and to recognise that conclusions are probabilistic, not absolute. In the OCR specification, you will encounter one-tailed and two-tailed tests, and you must be able to set up hypotheses correctly, calculate probabilities, and interpret results in context. Mastery of this topic is vital for the Statistics section of the exam, where it often appears in multi-step problems worth 8–12 marks.

This topic connects to other areas of A-Level Mathematics, such as probability, sampling, and data representation. Understanding hypothesis testing also prepares you for further study in statistics, economics, or any field that relies on data-driven decisions. In the exam, you will be expected to perform tests for a binomial proportion (using the binomial distribution) and for a normal mean (using the normal distribution, with known variance). You must also be comfortable with the language of significance, critical regions, and Type I and Type II errors.

• →Null and alternative hypotheses: H₀ represents the status quo or no effect; H₁ represents the claim you want to test. For a one-tailed test, H₁ specifies a direction (e.g., p > 0.5); for a two-tailed test, it does not (e.g., p ≠ 0.5).
• →Significance level (α): The probability of rejecting H₀ when it is true (Type I error). Common levels are 5% (0.05) and 1% (0.01). The critical region is the set of outcomes that would lead to rejection of H₀ at the chosen significance level.
• →Test statistic and p-value: The test statistic is calculated from sample data (e.g., number of successes in a binomial test). The p-value is the probability of obtaining a test statistic at least as extreme as the observed value, assuming H₀ is true. If the p-value ≤ α, reject H₀.
• →Critical region and critical value: For a given significance level, the critical region is the tail(s) of the distribution. The critical value is the boundary of the critical region. If the test statistic falls in the critical region, reject H₀.
• →Type I and Type II errors: Type I error is rejecting a true H₀ (probability = α). Type II error is not rejecting a false H₀ (probability = β). The power of a test is 1 – β.

• Correct formulation of null (H0) and alternative (H1) hypotheses using appropriate parameter notation.
• Clear statement of the significance level used for the test.
• Identification of the test statistic and comparison against critical values or p-values.
• Correct identification of 1-tail or 2-tail tests based on the alternative hypothesis.
• Conclusions must be stated in context, reflecting the probabilistic nature of the result (e.g., 'There is evidence at the 5% level to reject H0').
• Correct use of the acceptance and rejection regions.

• 💡Always define your parameters (e.g., 'let p be the population proportion') at the start of your hypothesis test.
• 💡Use the calculator functions for binomial and normal distributions to find probabilities or critical values efficiently.
• 💡Ensure your conclusion directly answers the question asked in the context of the scenario.
• 💡When using Pearson's correlation, ensure you use the provided table of critical values correctly.
• 💡Double-check if the test is 1-tail or 2-tail before determining the critical region.
• 💡Always define your hypotheses clearly in terms of the population parameter. Use the notation from the question (e.g., p for proportion, μ for mean). State the significance level and whether the test is one- or two-tailed. This sets up your answer for full marks.
• 💡When calculating probabilities for the binomial test, use the cumulative distribution function (CDF) on your calculator efficiently. For example, to find P(X ≤ 3) when X ~ B(20, 0.4), use the binomial CD function. Show the calculation step, but you don't need to write out all terms.
• 💡For the normal test, standardise the sample mean to a z-score: z = (x̄ - μ) / (σ/√n). Then compare to the critical value from the normal distribution table. Remember to check whether the population variance is known; if not, you may need to use a t-test (but OCR A-Level typically assumes known variance).

• Stating conclusions as absolute certainties (e.g., 'Waiting times have increased' instead of 'There is evidence to suggest...').
• Accepting the null hypothesis (the correct terminology is 'no evidence to reject H0').
• Incorrectly setting up 1-tail vs 2-tail tests.
• Failing to define the parameters used in the hypotheses (e.g., defining p as the population proportion).
• Misinterpreting the significance level as the probability of the null hypothesis being true.
• Misinterpreting the p-value: A common mistake is thinking the p-value is the probability that H₀ is true. In fact, it is the probability of observing the data (or more extreme) given that H₀ is true. The p-value does not directly tell you the truth of H₀.
• Confusing one-tailed and two-tailed tests: Students often forget to double the probability for a two-tailed test when using the binomial distribution. For a two-tailed test at 5% significance, you need to find the critical region such that the total probability in both tails is ≤ 0.05, with each tail having approximately half.
• Incorrectly stating hypotheses: H₀ and H₁ must be about a population parameter (e.g., p or μ), not about the sample. Also, H₁ must be the opposite of H₀. For example, if H₀: p = 0.3, then H₁ could be p > 0.3, p < 0.3, or p ≠ 0.3, but not p = 0.4.