University of Sydney · FACULTY OF STATISTICS

BUSS1020 Chap.8 One-Sample Hypothesis Tests

- one subject, every graph, every model, every mark
6 Chapters3-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 8 of 10 · BUSS1020

One-Sample Hypothesis Tests

A one-sample test compares sample evidence with a reference claim about one population parameter. This chapter builds null and alternative hypotheses, chooses one- or two-tailed directions before seeing results, and distinguishes Type I error, Type II error and power. It interprets p-values under the null model and uses disciplined conclusion language: reject or fail to reject rather than prove or accept.

Worked examples cover a one-mean t test and a one-proportion z test. The significance level is tied to decision error rather than treated as a magic number. A result near a conventional threshold is reported with full precision and an effect estimate, while a non-rejection remains evidence that was insufficient rather than proof of equality.

The ethics section keeps tail choice, subgroup choice and communication rules fixed before inspecting favourable results. Error-cost examples force truth and decision into the same sentence, reducing swapped Type I and Type II labels. The closing interpretation reports practical change beside statistical evidence so the threshold does not make the whole managerial decision.

Assumption checks remain beside the p-value, because selective sampling or dependence can invalidate precise-looking evidence.

In this chapter

What this chapter covers

  • 01

    Null and alternative hypotheses

  • 02

    Type I and Type II errors

  • 03

    Significance level and power

  • 04

    Critical-value and p-value approaches

  • 05

    One-sample tests for a mean

  • 06

    One-sample tests for a proportion

Worked example · free

Test a reduction in preparation time

Q [5 marks]. A random sample of 36 orders has mean preparation time 28.8 minutes and sample standard deviation 4.8 minutes. Test a population mean of 30 against the claim that the mean is lower. This is an AskSia-authored drill; the mark allocation shown here is illustrative and not an official university assessment scheme.
The testing chain starts with a parameter claim, compares sample evidence under the null model, and ends with a contextual conclusion rather than a claim of proThe testing chain starts with a parameter claim, compares sample evidence under the null model, and ends with a contextual conclusion rather than a claim of proof.hypothesestest statisticp-valueconclusionα is chosen before seeing the result
The testing chain starts with a parameter claim, compares sample evidence under the null model, and ends with a contextual conclusion rather than a claim of proof.
  • 1State H0: mu equals 30 and H1: mu is less than 30 before using the data.
  • 1Compute standard error as 4.8 divided by 6, giving 0.8 minutes.
  • 1Compute t as 28.8 minus 30 divided by 0.8, giving negative 1.5.
  • 1Use the left-tail p-value of about 0.071 and compare it with alpha 0.05.
  • 1Fail to reject and state insufficient evidence that the population mean is below 30 minutes.
The p-value is about 0.071, so the sample does not cross the five per cent threshold. This is insufficient evidence for the reduction claim, not proof that the mean equals 30.
Sia tip — Choose the tail from the business claim before seeing the sample direction; changing it afterwards understates error risk.
Glossary

Key terms

Null hypothesis
The reference population claim containing the equality used for the test calculation.
Alternative hypothesis
The directional or two-sided population claim the study seeks evidence for.
Type I error
Rejecting a null hypothesis that is actually true.
Type II error
Failing to reject a null hypothesis that is actually false at a specified alternative.
Power
The probability that a test rejects the null at a specified true alternative.
p-value
The null-model probability of evidence at least as inconsistent with the null as the observation.
FAQ

One-Sample Hypothesis Tests FAQ

Why is the null written with equality?

The equality supplies a precise reference value for the sampling distribution and test statistic. The alternative contains greater than, less than or not equal according to the business claim and determines which tail areas count as evidence.

Does failing to reject prove the null?

No. It means the sample did not provide enough evidence for the alternative at the chosen threshold. The study may have low power, high variability or limited sample size, so a real difference can remain undetected.

Is a small p-value practically important?

Not necessarily. A large sample can make a tiny effect statistically detectable. Report the estimated effect, an uncertainty interval, assumptions and business consequences so decision makers can judge practical importance rather than using a threshold alone.

Study strategy

Exam move

Write hypotheses in symbols and words, with the population parameter visible. Practise mapping right, left and two-sided business claims to tail areas before calculating. Create two-column error statements that name what is true and what the procedure decides. Rehearse one-mean and one-proportion tests with their different null standard errors.

For every p-value, say what is assumed and avoid calling it the probability that the null is true. Pair the decision with an estimate and a sentence about practical effect size whenever possible. Start a test by writing the claim that would change the decision and whether evidence in one or both directions matters. Set alpha before looking at the sample.

Build the statistic as estimate minus null value over its null standard error, and predict the sign from the sample direction. When the p-value is close to the threshold, keep the exact value and avoid storytelling around a rounded result. For a one-proportion test, use the null proportion in the test denominator and compare that with the sample-based denominator used by an interval.

Practise Type I and Type II errors using hiring, fraud and service examples so the labels remain tied to truth and decision. Add a power sentence that names effect size, variability and sample size. Close every test with reject or fail to reject, sufficient or insufficient evidence, the population parameter and an explicit warning that significance alone does not establish practical importance or causation.

Working through One-Sample Hypothesis Tests in BUSS1020? Sia is AskSia’s AI Statistics tutor — ask any BUSS1020 One-Sample Hypothesis Tests question and get a clear, step-by-step explanation grounded in how BUSS1020 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 202 of your subjects at University of Sydney, plus 1,000+ more Bibles across the AskSia library.
Sia - your BUSS1020 tutor, unlimited, worked the way the exam marks it
The full 3-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works