Did you know? A z-statistic of 1.80 can lead to a rejection at the 5% level in a valid one-tailed test, yet not in a two-tailed test. The data have not changed. The question you promised to test has changed.
This JC2 H2 Mathematics tutor guide for Bukit Timah readers answers a precise A-Level 9758 problem: How do you decide between a one-tailed and two-tailed hypothesis test? The decision is made from the claim and the alternative hypothesis, not from whichever test makes the answer significant.
Start with the claim in ordinary English, translate it into H₀ and H₁, and only then calculate a test statistic or a p-value. The distinction between a directional and a non-directional claim prevents one of the most avoidable errors in JC2 statistics.
Quick answer: the alternative hypothesis sets the tail
- “Has the average increased?” Use an upper-tail alternative: H₁: μ > μ₀.
- “Has the average decreased?” Use a lower-tail alternative: H₁: μ < μ₀.
- “Has the average changed?” Use a two-tailed alternative: H₁: μ ≠ μ₀.
- Before looking at the observed mean: fix the direction, significance level and population quantity being tested.
The null hypothesis expresses the reference value, usually H₀: μ = μ₀. The alternative specifies which departures would count as evidence. A one-tailed test devotes its rejection region to one prespecified direction; a two-tailed test considers departures in either direction.
Worked example: same z = 1.80, different claims
Imagine a fictional study of a numerical assessment where the claimed population mean is 50, population standard deviation is known to be 10, and an independent random sample of 100 observations has mean 51.8. Under H₀: μ = 50, the standard error is 10/√100 = 1. The test statistic is z = (51.8 − 50)/1 = 1.80. Use a 5% significance level.
Claim A: “The population mean has increased”
If that directional claim was set before inspecting the data, H₁: μ > 50. The upper-tail p-value is P(Z ≥ 1.80) ≈ 0.0359. Since 0.0359 < 0.05, reject H₀ at the 5% level. In context: there is statistical evidence, under the stated model, that the population mean exceeds 50.
Claim B: “The population mean has changed”
For this non-directional claim, H₁: μ ≠ 50. The two-tailed p-value is approximately 2 × 0.0359 = 0.0719. Since 0.0719 > 0.05, do not reject H₀ at the 5% level. The data are insufficient to establish a change using this test and threshold. That is not evidence that the mean must equal 50.
Same sample, different question. Do not inspect 51.8, decide that “increased” sounds more convenient and switch from a preplanned two-tailed test to a one-tailed test. This reverses the logic of hypothesis testing.
A decision tree students can use in seconds
Step 1: Underline the verb. Increased, improved, higher, exceeded suggest “greater than”. Reduced, lower, fell suggest “less than”. Changed or different suggests “not equal”. Step 2: Name the population parameter. Is the claim about μ or about a sample mean? Hypotheses refer to the population parameter. Step 3: State H₀ and H₁. Step 4: Choose α from the question. Step 5: Calculate and interpret in the original context.
A particularly useful check is to reverse the direction of the observed result. If H₁ is “μ > 50” but X̄ happens to be 48.2, the evidence points the wrong way for that claim. With the same known σ and n, z = −1.8 and the upper-tail p-value is approximately 0.9641, not 0.0359.
What the p-value actually says
A p-value is the probability, assuming the null model is true, of obtaining a result at least as extreme in the direction(s) specified by the test. It is not the probability that H₀ is true. A non-significant p-value is not a proof of H₀, and a significant p-value is not automatically an important real-world difference.
In the one-tailed example, the value 0.0359 represents the probability under H₀ of observing a sample mean at least 51.8 when n = 100 and σ = 10, within the stated normal-test model. In the two-tailed example, outcomes equally far or further on either side of 50 contribute to the probability.
Critical-region cross-check
For the standard normal reference distribution and α = 0.05, an upper-tail test rejects for approximately z > 1.645. A two-tailed test rejects when z is outside approximately −1.96 to 1.96. The observed 1.80 lies beyond 1.645 but inside the two-tailed non-rejection interval. The critical-region view and the p-value view should agree.
Do not use these z thresholds when the examination question prescribes another distribution or a different significance level. The 2026 H2 Mathematics syllabus addresses tests for a population mean in specified known-variance normal or large-sample settings, with contextual interpretation.
Five frequent mistakes in JC2 solutions
- Choosing the tail after seeing the data: fix H₁ from the research question first.
- Testing X̄ in the hypothesis: H₀ and H₁ are statements about the population mean μ.
- Doubling every p-value: two-tailed conversion depends on symmetry and the test being performed.
- Mixing up p and α: p comes from data under H₀; α is the preselected decision threshold.
- Writing “accept H₀” as certainty: prefer “do not reject H₀; insufficient evidence for the alternative”.
One-minute tutor check: speak before calculating
Present these claims, without sample data: “A new approach raises mean performance”; “mean performance differs”; “mean performance falls”. Ask the student to write H₁ each time. The correct signs, in order, are >, ≠, <. Then show the sample summary, ask the learner to compute z, and ask for a verbal conclusion in the language of the claim.
A strong learner also knows what the method does not say: that a statistical result guarantees an individual student will improve, or that statistical significance measures the educational importance of the effect.
Eight quick cases and answers
- 1. “Average time has decreased”: H₁ is μ < μ₀.
- 2. “Average time is different”: H₁ is μ ≠ μ₀.
- 3. With z = 1.80, upper-tail p ≈ 0.0359.
- 4. With z = 1.80, two-tailed normal p ≈ 0.0719.
- 5. For α = 0.01 and p = 0.0359, do not reject H₀.
- 6. For H₁: μ > μ₀ and z = −1.80, the upper-tail p ≈ 0.9641.
- 7. Known σ = 10, n = 25: standard error is 2.
- 8. Mean 53.6 versus μ₀ = 50, σ = 10, n = 25: z = 1.80.
Where this sits in the syllabus
Hypothesis testing is part of H2 Mathematics 9758 Probability and Statistics. The official syllabus includes H₀/H₁, test statistic, p-value, critical region, significance level, one- and two-tailed tests, and interpretation. It specifies tests for one population mean with the applicable known-variance or large-sample setup. It explicitly excludes testing the difference between two population means; do not import that more advanced method into a routine 9758 answer.
For the broad concept and more scenarios, visit H2 Mathematics Hypothesis Testing and p-Values. This page deals only with setting the right direction before the calculation.
Frequently asked questions
Can I change a two-tailed test into a one-tailed test because my sample mean increased?
Not as a post-hoc way to obtain a smaller p-value. A directional alternative needs justification in the question or a prespecified investigative aim.
Does p > 0.05 mean H₀ is true?
No. It means the result did not cross the chosen decision threshold for the specified test. The evidence could be limited for several reasons.
Is a p-value the chance that the null hypothesis is true?
No. It is calculated as a probability of sample evidence assuming the null model, not a probability assigned to the truth of H₀.
What if the question says “changed” but the sample mean is larger?
Changed still indicates a two-tailed alternative unless the actual task supplies a justified directional hypothesis.
How should the last sentence be written?
State the significance level, reject or do not reject H₀, and describe whether there is sufficient evidence for the precise population claim in the question.
Continue learning: four distinct JC2 H2 questions
- Why Standard Error Falls by √n
- One-Tailed or Two-Tailed Hypothesis Test? — this guide
- Which Regression Line Predicts Correctly?
- Line–Plane Angle vs Normal Vector
For the wider subject route, read the Bukit Timah JC H2 Mathematics guide and the Bukit Timah JC2 learning hub. These are subject and level overviews; this article owns only the specific question explained above.
Syllabus and scope: SEAB Singapore–Cambridge H2 Mathematics (9758), 2026 examination syllabus. The numbers in worked questions here are invented teaching examples, not past examination questions or actual student records. Refer to the syllabus for your examination year.