H2 Maths Hypothesis Tests: Choose the Tail Before Calculating

Use the claim to choose the direction

In an SGExams H2 Maths discussion, students debated whether a hypothesis test should use a left tail and how a critical region related to a p-value. Those are questions about the test setup, not just calculator commands. Work out the alternative hypothesis before looking at whether the sample mean happens to be high or low.

The 2026 H2 Mathematics 9758 syllabus includes one- and two-tailed tests for a population mean, p-values and interpretation in context. Check the stated assumptions and your own examination-year syllabus. This guide uses an original normal-population example and does not reconstruct the 2023 exam question.

Define the parameter and the hypotheses

Suppose a machine is claimed to fill containers with a population mean of 500 ml. If you are testing whether the mean is lower, write H0: μ = 500 and H1: μ < 500, where μ is the population mean fill volume in millilitres. If the concern is any change, the alternative is μ ≠ 500.

Do not choose a lower-tail test only because the observed sample mean is below 500. The concern being tested determines the direction. State it before calculating. Also distinguish the population parameter μ from the observed sample mean, which is a statistic calculated from your sample.

Standardise with the standard error

For an invented example, assume independent observations from a normal population with known standard deviation 4 ml. With sample size 25 and observed mean 498.8 ml, the standard error is 4/√25 = 0.8 ml. The standardised statistic is (498.8 − 500)/0.8 = −1.5.

For the lower-tail alternative, the p-value is approximately 0.0668. At a 5% significance level, this exceeds 0.05, so do not reject H0. The probability is calculated under the null model. It is not a 6.68% probability that H0 itself is true, nor a probability that the observed mean will recur exactly.

Change the question and check the tail

If the alternative were μ ≠ 500 in the same normal example, outcomes equally far in either direction would count against H0. The two-tailed p-value is therefore about 0.1336. The calculation changes because the alternative changed, not because the numerical statistic became different.

This doubling works here because the null distribution is continuous and symmetric and the two-tailed criterion uses equal extremeness in either direction. Do not apply an unexplained doubling habit to every probability problem. If a question asks for a critical region instead, use the specified method and check that the rejection decision is consistent.

Write what the test supports

For the lower-tail example, conclude that there is insufficient evidence at the 5% level that the population mean fill is below 500 ml. Do not say the test proves the machine has the claimed mean. Review a marked question by separating the hypotheses, assumptions, statistic, probability and contextual conclusion so you can identify the precise step that failed.

Choose a suitable practice question from our H2 Maths papers, checking the level, year and original instructions.

Sources and further reading

Checked on 6 November 2026. The linked discussion is real; worked examples and practice routines are our original guidance. Older discussions are not evidence of current examination rules.

• A Levels 9758/02 H2 Mathematics Megathread — 10 November 2023.
• SEAB 2026 syllabus