Loading subjects
Evaluating Statistical Reports · Part 2 of 5
9 exam-style questions with model answers, plus 12 quick multi-choice questions — every question on this part of the standard, grouped by the 3 pages of notes they come from.
Write a full answer before you reveal the model one. That comparison is where the learning happens.
Describe stratified sampling and give one reason a researcher might choose it over simple random sampling.
A magazine prints a questionnaire and invites readers to post it back. 4 200 replies are received and the magazine reports that '78% of readers are dissatisfied with their broadband provider'. Explain why this figure is likely to overstate dissatisfaction, referring to the sampling process.
A market research firm reports: 'We recruited 1 500 respondents from our online panel and weighted results by age, gender and region so the sample matches the New Zealand population. Our margin of error is 2.5%. 61% of New Zealanders say they would pay more for locally produced food.' Assess this methodology with respect to the conclusion drawn.
Explain the difference between undercoverage and non-response bias, with an example of each.
A council survey asks: 'Do you support the council's plan to protect our parks by increasing rates by 2%?' 68% said yes. Explain two problems with this question and what each means for the 68% figure.
A government agency reports: 'A face-to-face survey of 2 000 randomly selected adults found that 12% reported drinking more than the recommended weekly alcohol limit, down from 18% in the previous survey, which was conducted online. Hazardous drinking has fallen sharply.' Assess this report with respect to its conclusion.
A poll of 1 000 voters has a maximum margin of error of about 3.1%. A party's support moves from 24% to 26%. What can be concluded?
A poll reports party percentages after excluding the 16% of respondents who were undecided. Explain what this means for how the published figures should be interpreted, and one risk it creates.
Two polls are published in the same week. Company A (n = 1 000, phone) has Party X on 34%; Company B (n = 1 200, online panel) has Party X on 29%. A commentator writes: 'The polls disagree, so at least one of them must be wrong.' Assess this statement.