Pilot Surveys (Higher) - Worksheets, Questions and Revision
59 original exam-style questions - 26 pages of questions with a full mark scheme - free printable PDF.
Pilot Surveys (Higher) revision guideMore GCSE Statistics practiceMock exams
This topic is chapter 4 of GCSE Statistics Higher Topic Workbook 1 and chapter 4 of GCSE Statistics Higher 10-Minute Tests 1.
1.8 Pilot Surveys (Higher)
Give four different reasons why the company might do this.
- A) Collect
- B) Plan
- C) Process
- D) Interpret
Question 1: "Did you enjoy your visit today?" Response options: Yes / No
Question 2: "What could we do to improve your visit?" (visitors write their own answer, with no response options given)
| Year group | Number of students |
|---|---|
| Year 12 | 450 |
| Year 13 | 300 |
Before running a full survey about a proposed new common room, the college pilots its questionnaire using a stratified sample of 50 students.
Design A: give the questionnaire to 15 staff members' friends and family to complete.
Design B: select a random sample of 200 loyalty-card holders from across the country and post the questionnaire to them.
Compare Design A and Design B, and explain which would give a more useful pilot for testing the questionnaire before the main national survey.
Discuss the benefits and limitations of using a pilot survey in this situation.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 100 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 300 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 300 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Explain why this is a poor question.
(b) Suggest how to fix it.
| Pair | Without method | With method |
|---|---|---|
| Pair 1 | 73 | 83 |
| Pair 2 | 47 | 55 |
| Pair 3 | 48 | 53 |
| Pair 4 | 90 | 91 |
| Pair 5 | 77 | 76 |
| Pair 6 | 70 | 68 |
(a) Work out the difference (with minus without) for each pair.
(b) Work out the mean difference. Give your answer to 2 decimal places if needed.
31, 27, 52, 28, 29, 30
(a) Write down the value that looks anomalous.
(b) Work out the mean of the remaining 5 values. Give your answer to 1 decimal place if needed.
(c) Give one reason for repeating an experiment.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 150 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Give one reason why many people may not answer this.
(b) Suggest how the question could be changed to make people more willing to answer.
(a) Work out the response rate as a percentage, to 1 decimal place if needed.
(b) The main survey needs 400 returned questionnaires. Estimate how many must be sent out if the response rate stays the same. Round up to a whole number.
(a) How many volunteers are in each group?
(b) Give one reason for using a placebo.
(c) Describe how the volunteers could be allocated to groups at random.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 300 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Explain what the problem suggests about the questionnaire.
(b) State what should be done before the main survey.
(a) Explain what is wrong with these response boxes, giving a value that could go in two boxes.
(b) Write down a better set of three response boxes starting from 1.
| Pair | Without method | With method |
|---|---|---|
| Pair 1 | 41 | 42 |
| Pair 2 | 55 | 61 |
| Pair 3 | 41 | 53 |
| Pair 4 | 76 | 74 |
| Pair 5 | 52 | 63 |
| Pair 6 | 65 | 61 |
| Pair 7 | 42 | 41 |
(a) Work out the difference (with minus without) for each pair.
(b) Work out the mean difference. Give your answer to 2 decimal places if needed.
(a) Explain what is wrong with these response boxes, giving a value that could go in two boxes.
(b) Write down a better set of three response boxes starting from 1.
| Pair | Without method | With method |
|---|---|---|
| Pair 1 | 46 | 46 |
| Pair 2 | 41 | 52 |
| Pair 3 | 74 | 77 |
| Pair 4 | 67 | 65 |
| Pair 5 | 83 | 83 |
(a) Work out the difference (with minus without) for each pair.
(b) Work out the mean difference. Give your answer to 2 decimal places if needed.
(a) Explain what the problem suggests about the questionnaire.
(b) State what should be done before the main survey.
27, 4, 31, 31, 31
(a) Write down the value that looks anomalous.
(b) Work out the mean of the remaining 4 values. Give your answer to 1 decimal place if needed.
(c) Give one reason for repeating an experiment.
(a) Explain what the problem suggests about the questionnaire.
(b) State what should be done before the main survey.
"Do you agree that bus services is clearly excellent and that everyone should support it?"
(a) Give one reason why this is a poor question.
(b) Write a better question.
(a) How many volunteers are in each group?
(b) Give one reason for using a placebo.
(c) Describe how the volunteers could be allocated to groups at random.
(a) Explain what is wrong with these response boxes, giving a value that could go in two boxes.
(b) Write down a better set of three response boxes starting from 1.
48, 52, 49, 26, 51, 54
(a) Write down the value that looks anomalous.
(b) Work out the mean of the remaining 5 values. Give your answer to 1 decimal place if needed.
(c) Give one reason for repeating an experiment.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 200 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
45, 42, 66, 44, 41, 42
(a) Write down the value that looks anomalous.
(b) Work out the mean of the remaining 5 values. Give your answer to 1 decimal place if needed.
(c) Give one reason for repeating an experiment.
| Pair | Without method | With method |
|---|---|---|
| Pair 1 | 43 | 52 |
| Pair 2 | 83 | 79 |
| Pair 3 | 62 | 74 |
| Pair 4 | 41 | 39 |
| Pair 5 | 79 | 83 |
(a) Work out the difference (with minus without) for each pair.
(b) Work out the mean difference. Give your answer to 2 decimal places if needed.
(a) Give one reason why many people may not answer this.
(b) Suggest how the question could be changed to make people more willing to answer.
44, 40, 16, 38, 39, 43
(a) Write down the value that looks anomalous.
(b) Work out the mean of the remaining 5 values. Give your answer to 1 decimal place if needed.
(c) Give one reason for repeating an experiment.
(a) Work out the response rate as a percentage, to 1 decimal place if needed.
(b) The main survey needs 400 returned questionnaires. Estimate how many must be sent out if the response rate stays the same. Round up to a whole number.
(a) How many volunteers are in each group?
(b) Give one reason for using a placebo.
(c) Describe how the volunteers could be allocated to groups at random.
(a) Explain what the problem suggests about the questionnaire.
(b) State what should be done before the main survey.
(a) Explain what is wrong with these response boxes, giving a value that could go in two boxes.
(b) Write down a better set of three response boxes starting from 2.
(a) Work out the response rate as a percentage, to 1 decimal place if needed.
(b) The main survey needs 1000 returned questionnaires. Estimate how many must be sent out if the response rate stays the same. Round up to a whole number.
(a) Give one reason why many people may not answer this.
(b) Suggest how the question could be changed to make people more willing to answer.
(a) How many volunteers are in each group?
(b) Give one reason for using a placebo.
(c) Describe how the volunteers could be allocated to groups at random.
(a) Work out the mean time per person.
(b) Estimate the total time needed to survey 100 people, in minutes.
(c) Give one reason for carrying out a pilot survey.
(a) Work out the response rate as a percentage, to 1 decimal place if needed.
(b) The main survey needs 800 returned questionnaires. Estimate how many must be sent out if the response rate stays the same. Round up to a whole number.
"Do you agree that school lunches is clearly excellent and that everyone should support it?"
(a) Give one reason why this is a poor question.
(b) Write a better question.
(a) Give one reason why many people may not answer this.
(b) Suggest how the question could be changed to make people more willing to answer.
(a) Explain what is wrong with these response boxes, giving a value that could go in two boxes.
(b) Write down a better set of three response boxes starting from 0.
Recommend, with reasons, what the airport should do at the plan stage and the collect stage before running its main survey of all passengers.
Evaluate this claim.
Question 1
- (a) a small-scale trial run of a survey or questionnaire, carried out before the main survey, oeB1
- (a) Answer: A pilot survey is a small-scale trial run of a survey or data collection method, carried out on a small group before the main survey takes place.
- (b) any valid purpose, e.g. checks the wording of questions is clear, oe checks how long the survey takes, oe estimates the likely response rateB1
- (b) Answer: To check that the questions are clear and easy for people to understand before the main survey.
- (c) the pilot uses a much smaller sample size than the main survey, oe it happens earlier/before the main surveyB1
- (c) Answer: A pilot survey is carried out on a much smaller number of people than the main survey.
Question 2
- (i) True caoB1
- (i) Answer: True
- (ii) False caoB1
- (ii) Answer: False
- (iii) True caoB1
- (iii) Answer: True
Question 3
- (a) checks that the wording of the questions about the app is clear and not confusing for passengers, oeB1
- (a) Answer: To check that the questions about the app are worded clearly and are not confusing.
- (b) checks how long the questionnaire takes passengers to complete, oeB1
- (b) Answer: To find out how long the questionnaire takes passengers to complete.
- (c) gives an estimate of the likely response rate, so the company can plan how many passengers to contact for the main survey, oeB1
- (c) Answer: To estimate the response rate, so the company knows roughly how many passengers to contact in the main survey.
- (d) checks that the method of collecting responses (e.g. an app link or paper form) works properly in practice before it is used on a large scale, oeB1
- (d) Answer: To check that the practical method of collecting responses, such as a link sent through the app, actually works before it is used nationwide.
Question 4
- (a) identifies that all 8 pilot respondents share the same course, so are not representative of the whole student populationB1
- (a) explains that problems found (or not found) in this pilot might not reflect problems that would occur with the wider student population, e.g. students on other courses may have different travel patterns/viewsB1
- (a) Answer: All 8 respondents are on her own course, so they are not representative of the whole student population; problems that would affect students on other courses might not show up in this pilot.
- (b) suggests selecting students from a range of courses, year groups or halls of residence, e.g. a random or stratified sampleB1
- (b) explains this would make the pilot sample more representative of the population being surveyed in the main surveyB1
- (b) Answer: She should select pilot respondents from a range of different courses and year groups, for example using a random sample, so the pilot is more representative of the whole student population.
Question 5
- (a) it is a leading question caoB1
- (a) Answer: It is a leading question.
- (b) it pushes/encourages members towards agreeing, so responses would be biased and not reflect members' true opinions, oeB1
- (b) Answer: The wording pushes members towards agreeing, so the results would be biased and would not show what members really think.
- (c) no longer contains wording that suggests an expected answerB1
- (c) remains clear and answerable, e.g. offers a genuine scale of responseB1
- (c) Answer: e.g. "What do you think of the new gym equipment?" with response options Much better, Better, About the same, Worse, Much worse.
Question 6
- (a) 51/60, oeM1
- (a) 85% caoA1
- (a) Answer: 85%
- (b) high, with an acceptable reason, e.g. most people who received a questionnaire returned it (51 out of 60)B1
- (b) Answer: High. Most of the customers who were sent a questionnaire (51 out of 60) returned it.
Question 7
- (a) e.g. weather conditions on the day (wind, rain or temperature) affect how many ladybirds are active/visible, so counts vary for reasons unrelated to the true population, oeB1
- (a) Answer: Weather conditions on the day, such as wind or rain, affect how many ladybirds are active and visible, so the count could vary for reasons that have nothing to do with the true population size.
- (b) e.g. the sweep net may be difficult to use accurately in the time allowed, or small ladybird larvae may be hard to identify and count quickly, oeB1
- (b) Answer: The time allowed may not be long enough to sweep the whole quadrat accurately, or small ladybird larvae may be hard to identify and count quickly enough.
- (c) problems found can be corrected (e.g. choosing calmer days, allowing more time, using clearer identification guides) before the main data collection takes placeB1
- (c) this makes the counts from the main survey more consistent/reliable and comparable between different meadows, oeB1
- (c) Answer: Problems found in the pilot, such as poor weather or rushed identification, can be fixed before the main survey, for example by choosing calmer days or allowing more time. This makes the counts from the main survey more consistent and fair to compare between meadows.
Question 8
- (a) any valid reason, e.g. the email was missed/went to a spam folder, oe patients did not have time, oe patients were not comfortable giving feedback onlineB1
- (a) Answer: The email might have gone to a spam folder, or patients simply did not have time to complete it.
- (b) identifies that too few responses would likely be collected overall if the same low rate continued at a larger scaleB1
- (b) identifies that patients who do respond might not be typical of all patients (e.g. only those with strong opinions reply), so conclusions from the main survey could be biasedB1
- (b) Answer: Too few patients would respond to give a reliable sample size in the main survey, and the small number who do respond might not be typical of all patients, for example only those with strong opinions might reply, so the conclusions could be biased.
Question 9
- (a) B caoB1
- (a) Answer: B) Plan
- (b) identifies that the confusing question should be reworded/clarified at the plan stage before data collection beginsB1
- (b) explains that the collect stage will then use the improved/corrected questionnaire on the full membership, producing clearer, more valid dataB1
- (b) Answer: The confusing question should be reworded and clarified before the collect stage starts. The collect stage will then use the improved questionnaire on all members, producing clearer, more valid responses than if the original wording had been used.
Question 10
- (a) suggests changing the closed Yes/No question to one with a wider range of options, e.g. a rating scale such as Excellent, Good, Average, Poor, or adding a follow-up question asking visitors to explain their answerB1
- (a) explains this would reveal more detail about visitors' opinions than a simple Yes/No response, oeB1
- (a) Answer: Replace the Yes/No question with a rating scale (e.g. Excellent, Good, Average, Poor), or add a follow-up asking visitors to explain their answer, so the museum learns more detail than a simple Yes/No gives.
- (b) suggests changing the open question to a closed question with a set of response categories/tick boxes covering common areas, e.g. signage, cafe, exhibits, staffB1
- (b) explains fixed categories are quicker/easier for staff to count and compare across all visitors than free-text answers, oeB1
- (b) Answer: Give visitors a closed list of tick-box categories (e.g. signage, cafe, exhibits, staff) instead of a blank space, because fixed categories are much easier for staff to count and compare across every visitor than free-text answers.
Question 11
- (a) the response categories overlapB1
- (a) £40,000 appears in both the '£20,000-£40,000' box and the '£40,000-£60,000' box, so it is unclear which one to tickB1
- (a) Answer: The categories overlap. A household income of exactly £40,000 could be ticked in either the '£20,000-£40,000' box or the '£40,000-£60,000' box, so it is not clear which one to choose.
- (b) boundaries no longer shared between adjacent categories, e.g. Under £20,000, £20,000-£39,999, £40,000-£59,999B1
- (b) categories cover every income with no gaps, including an open-ended top category, e.g. £60,000 and overB1
- (b) Answer: e.g. Under £20,000, £20,000-£39,999, £40,000-£59,999, £60,000 and over
Question 12
- (a) gives a reasonable percentage, e.g. between 5% and 10% (accept any sensible value in this range), oeB1
- (a) justifies the choice: large enough to reveal problems with the questionnaire and give a reasonable estimate of the response rate, but small enough not to use up too much of the time/budget intended for the main surveyB1
- (a) Answer: e.g. Around 5% to 10% of the main sample. This is large enough to reveal wording or logistical problems and to estimate the likely response rate, but small enough to avoid using up too much of the time and budget meant for the main survey.
- (b) 8000 x 0.05, oeM1
- (b) 400 caoA1
- (b) Answer: 400
- (c) with so few responses, any problems found (or the response rate estimate) could occur by chance and might not reliably represent what would happen with the full 8000 customers, oeB1
- (c) Answer: With only 10 customers, any problems found (or not found) could just be due to chance, and might not reliably predict what happens across all 8000 customers.
Question 13
- (a) 450/750 x 50, oe (or equivalent ratio method)M1
- (a) 30 cao (Year 12)A1
- (a) 20 cao (Year 13)A1
- (a) Answer: 30 Year 12 students and 20 Year 13 students
- (b) (27 + 12)/50, oeM1
- (b) 78% caoA1
- (b) Answer: 78%
Question 14
- (a) identifies Question 1 as a leading question, e.g. 'caring, attentive' and 'Don't you agree' suggest the expected answerB1
- (a) gives an improved, unbiased version, e.g. 'What did you think of the nurses on this ward?' with a genuine scale of response optionsB1
- (a) Answer: Question 1 is leading, because 'caring, attentive' and 'Don't you agree' suggest the expected answer. Improved: 'What did you think of the nurses on this ward?' with options such as Excellent, Good, Average, Poor.
- (b) identifies that Question 2 asks about two different things at once (doctors and food), so a patient who is happy with one but not the other cannot answer clearlyB1
- (b) rewrites as two separate questions, e.g. 'What did you think of the doctors?' and 'What did you think of the food?'B1
- (b) Answer: Question 2 asks about two different things (doctors and food) at once, so a patient satisfied with one but not the other cannot give one clear answer. Rewrite as: 'What did you think of the doctors?' and 'What did you think of the food?'
- (c) identifies that the waiting-time categories overlap, e.g. 15 minutes appears in both '0-15 minutes' and '15-30 minutes', and 30 minutes appears in both '15-30 minutes' and '30-60 minutes'B1
- (c) gives improved, non-overlapping categories that still cover every waiting time, e.g. 0-14 minutes, 15-29 minutes, 30-59 minutes, 1 hour or moreB1
- (c) Answer: The categories overlap at 15 and 30 minutes. Improved: 0-14 minutes, 15-29 minutes, 30-59 minutes, 1 hour or more.
Question 15
- an overall judgement that Design B is more useful/appropriateB1
- explains that Design A's sample (staff friends/family) is unlikely to be representative of loyalty-card holders nationally, e.g. they may share similar ages, locations or views to staff, so problems affecting the wider population might not be revealedB1
- explains that Design B's random national sample is more representative of the population targeted by the main survey, so issues found (e.g. with wording or response rate) are more likely to also apply to the main surveyB1
- notes that Design B's larger sample size (200 compared with 15) also gives a more reliable early estimate of things such as response rate, before the expense of the full national rolloutB1
- Answer: Design B is more useful. A random national sample of 200 loyalty-card holders is far more representative of the people who will receive the main survey than 15 staff friends and family, who are likely to share similar characteristics to staff. Design B's larger, representative sample also gives a more reliable early estimate of the response rate before the costly national rollout.
Question 16
- a clear overall judgement is given (e.g. the pilot is worthwhile, even though it has some limitations), rather than only listing pointsB1
- a benefit is explained, e.g. testing on 150 customers is quick and low-cost, and can reveal problems with the questionnaire before it is sent to all 2,000,000 customersB1
- a limitation is explained, e.g. 150 is a tiny fraction of 2,000,000 customers (0.0075%), so the pilot might not reveal rarer problems that only affect a small number of customers, such as those with unusual tariffs or devicesB1
- Answer: Running a pilot survey is worthwhile. Testing on 150 customers is quick and low-cost, and lets the phone network fix problems with the questionnaire before spending time and money contacting all 2,000,000 customers. However, 150 is a tiny fraction (0.0075%) of 2,000,000, so rarer problems, such as issues affecting customers with unusual tariffs or devices, might not be revealed by the pilot.
Question 17
- 90 / 12M1
- 7.5 minutesA1
- 750 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 7.5 minutes; (b) 750 minutes; (c) to test the questions before the full survey
Question 18
- 120 / 5M1
- 24 minutesA1
- 7200 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 24 minutes; (b) 7200 minutes; (c) to test the questions before the full survey
Question 19
- 80 / 12M1
- 6.67 minutesA1
- 2000 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 6.67 minutes; (b) 2000 minutes; (c) to test the questions before the full survey
Question 20
- it is a double question (asks two things at once, so the answer is unclear)B1
- ask two separate questions, one about cost and one about qualityB1
- Answer: (a) double-barrelled question; (b) split into two separate questions
Question 21
- with minus without for at least half the pairsM1
- differences 10, 8, 5, 1, -1, -2 caoA1
- sum of differences divided by 6M1
- 3.5 caoA1
- Answer: (a) 10, 8, 5, 1, -1, -2; (b) 3.5
Question 22
- 52B1
- sum of the 5 remaining values divided by 5M1
- 29 caoA1
- to improve reliability / to spot anomalies / to reduce the effect of random errorB1
- Answer: (a) 52; (b) 29; (c) to improve reliability
Question 23
- 60 / 12M1
- 5 minutesA1
- 750 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 5 minutes; (b) 750 minutes; (c) to test the questions before the full survey
Question 24
- personal or sensitive question, so people may refuse or give false answersB1
- use response boxes with ranges (e.g. under 20 000, 20 000 to 40 000, over 40 000) or make the survey anonymousB1
- Answer: (a) the question is personal and sensitive; (b) use income ranges or make the survey anonymous
Question 25
- 17/25 x 100M1
- 68% caoA1
- 400 / (17/25) or 400 x 25/17M1
- 589 caoA1
- Answer: (a) 68%; (b) 589
Question 26
- 30 in each groupB1
- so the effect of believing in the treatment is the same for both groups (acts as a control)B1
- e.g. give each volunteer a number and use random numbers or draw names from a hat for the treatment groupB1
- Answer: (a) 30; (b) a placebo acts as a control; (c) use random numbers or draw names to allocate
Question 27
- 80 / 8M1
- 10 minutesA1
- 3000 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 10 minutes; (b) 3000 minutes; (c) to test the questions before the full survey
Question 28
- identifies the likely issue: the wording is vagueB1
- changes the questionnaire: give numerical response boxesB1
- Answer: the wording is vague, so give numerical response boxes
Question 29
- the boxes overlap, e.g. 6 fits in two boxesB1
- a set of non-overlapping classes with no gapsM1
- e.g. 1 to 5, 6 to 10, 11 to 15 (or inequalities)A1
- Answer: (a) the boxes overlap, for example 6 fits in two boxes; (b) non-overlapping boxes such as 1 to 5, 6 to 10, 11 to 15
Question 30
- with minus without for at least half the pairsM1
- differences 1, 6, 12, -2, 11, -4, -1 caoA1
- sum of differences divided by 7M1
- 3.29 caoA1
- Answer: (a) 1, 6, 12, -2, 11, -4, -1; (b) 3.29
Question 31
- the boxes overlap, e.g. 4 fits in two boxesB1
- a set of non-overlapping classes with no gapsM1
- e.g. 1 to 3, 4 to 6, 7 to 9 (or inequalities)A1
- Answer: (a) the boxes overlap, for example 4 fits in two boxes; (b) non-overlapping boxes such as 1 to 3, 4 to 6, 7 to 9
Question 32
- with minus without for at least half the pairsM1
- differences 0, 11, 3, -2, 0 caoA1
- sum of differences divided by 5M1
- 2.4 caoA1
- Answer: (a) 0, 11, 3, -2, 0; (b) 2.4
Question 33
- identifies the likely issue: the wording is vagueB1
- changes the questionnaire: give numerical response boxesB1
- Answer: the wording is vague, so give numerical response boxes
Question 34
- 4B1
- sum of the 4 remaining values divided by 4M1
- 30 caoA1
- to improve reliability / to spot anomalies / to reduce the effect of random errorB1
- Answer: (a) 4; (b) 30; (c) to improve reliability
Question 35
- identifies the likely issue: the question may be unclearB1
- changes the questionnaire: reword itB1
- Answer: the question may be unclear, sensitive or badly worded, so reword it
Question 36
- it is a leading question (it pushes the respondent towards one answer)B1
- a neutral wording with no persuasionM1
- includes a range of response options (closed) or is clearly open and unbiasedA1
- Answer: (a) leading question; (b) e.g. What is your opinion of bus services? (very good, good, neutral, poor, very poor)
Question 37
- 18 in each groupB1
- so the effect of believing in the treatment is the same for both groups (acts as a control)B1
- e.g. give each volunteer a number and use random numbers or draw names from a hat for the treatment groupB1
- Answer: (a) 18; (b) a placebo acts as a control; (c) use random numbers or draw names to allocate
Question 38
- the boxes overlap, e.g. 3 fits in two boxesB1
- a set of non-overlapping classes with no gapsM1
- e.g. 1 to 2, 3 to 4, 5 to 6 (or inequalities)A1
- Answer: (a) the boxes overlap, for example 3 fits in two boxes; (b) non-overlapping boxes such as 1 to 2, 3 to 4, 5 to 6
Question 39
- 26B1
- sum of the 5 remaining values divided by 5M1
- 50.8 caoA1
- to improve reliability / to spot anomalies / to reduce the effect of random errorB1
- Answer: (a) 26; (b) 50.8; (c) to improve reliability
Question 40
- 90 / 5M1
- 18 minutesA1
- 3600 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 18 minutes; (b) 3600 minutes; (c) to test the questions before the full survey
Question 41
- 66B1
- sum of the 5 remaining values divided by 5M1
- 42.8 caoA1
- to improve reliability / to spot anomalies / to reduce the effect of random errorB1
- Answer: (a) 66; (b) 42.8; (c) to improve reliability
Question 42
- with minus without for at least half the pairsM1
- differences 9, -4, 12, -2, 4 caoA1
- sum of differences divided by 5M1
- 3.8 caoA1
- Answer: (a) 9, -4, 12, -2, 4; (b) 3.8
Question 43
- personal or sensitive question, so people may refuse or give false answersB1
- use response boxes with ranges (e.g. under 20 000, 20 000 to 40 000, over 40 000) or make the survey anonymousB1
- Answer: (a) the question is personal and sensitive; (b) use income ranges or make the survey anonymous
Question 44
- 16B1
- sum of the 5 remaining values divided by 5M1
- 40.8 caoA1
- to improve reliability / to spot anomalies / to reduce the effect of random errorB1
- Answer: (a) 16; (b) 40.8; (c) to improve reliability
Question 45
- 5/20 x 100M1
- 25% caoA1
- 400 / (1/4) or 400 x 20/5M1
- 1600 caoA1
- Answer: (a) 25%; (b) 1600
Question 46
- 20 in each groupB1
- so the effect of believing in the treatment is the same for both groups (acts as a control)B1
- e.g. give each volunteer a number and use random numbers or draw names from a hat for the treatment groupB1
- Answer: (a) 20; (b) a placebo acts as a control; (c) use random numbers or draw names to allocate
Question 47
- identifies the likely issue: the language is too complexB1
- changes the questionnaire: use simpler wordingB1
- Answer: the language is too complex, so use simpler wording
Question 48
- the boxes overlap, e.g. 4 fits in two boxesB1
- a set of non-overlapping classes with no gapsM1
- e.g. 2 to 3, 4 to 5, 6 to 7 (or inequalities)A1
- Answer: (a) the boxes overlap, for example 4 fits in two boxes; (b) non-overlapping boxes such as 2 to 3, 4 to 5, 6 to 7
Question 49
- 28/40 x 100M1
- 70% caoA1
- 1000 / (7/10) or 1000 x 40/28M1
- 1429 caoA1
- Answer: (a) 70%; (b) 1429
Question 50
- personal or sensitive question, so people may refuse or give false answersB1
- use response boxes with ranges (e.g. under 20 000, 20 000 to 40 000, over 40 000) or make the survey anonymousB1
- Answer: (a) the question is personal and sensitive; (b) use income ranges or make the survey anonymous
Question 51
- 12 in each groupB1
- so the effect of believing in the treatment is the same for both groups (acts as a control)B1
- e.g. give each volunteer a number and use random numbers or draw names from a hat for the treatment groupB1
- Answer: (a) 12; (b) a placebo acts as a control; (c) use random numbers or draw names to allocate
Question 52
- 120 / 12M1
- 10 minutesA1
- 1000 minutesB1
- to test the questions / check they are understood / estimate time and cost before the full surveyB1
- Answer: (a) 10 minutes; (b) 1000 minutes; (c) to test the questions before the full survey
Question 53
- 18/20 x 100M1
- 90% caoA1
- 800 / (9/10) or 800 x 20/18M1
- 889 caoA1
- Answer: (a) 90%; (b) 889
Question 54
- it is a leading question (it pushes the respondent towards one answer)B1
- a neutral wording with no persuasionM1
- includes a range of response options (closed) or is clearly open and unbiasedA1
- Answer: (a) leading question; (b) e.g. What is your opinion of school lunches? (very good, good, neutral, poor, very poor)
Question 55
- personal or sensitive question, so people may refuse or give false answersB1
- use response boxes with ranges (e.g. under 20 000, 20 000 to 40 000, over 40 000) or make the survey anonymousB1
- Answer: (a) the question is personal and sensitive; (b) use income ranges or make the survey anonymous
Question 56
- the boxes overlap, e.g. 3 fits in two boxesB1
- a set of non-overlapping classes with no gapsM1
- e.g. 0 to 2, 3 to 5, 6 to 8 (or inequalities)A1
- Answer: (a) the boxes overlap, for example 3 fits in two boxes; (b) non-overlapping boxes such as 0 to 2, 3 to 5, 6 to 8
Question 57
- a clear overall recommendation is given, e.g. revise both the questionnaire and the sampling method before proceeding to the main surveyB1
- plan-stage point: reword the leading question identified in the pilot so it no longer suggests an expected answerB1
- a different plan-stage point: redesign the sampling method so future samples are drawn from passengers on a range of flights, times and destinations, not just one early-morning flightB1
- collect-stage point: take steps to improve the response rate for the main survey (e.g. sending reminder emails, or offering a paper option at the gate), since a 25% pilot response rate is lowB1
- a different collect-stage point: use a broader/more representative method of contacting passengers for the main data collection, so as not to under-represent passengers on other flights or those less likely to check emailB1
- explicitly links recommendations to the pilot evidence given (the 25% figure and the single-flight sample), rather than giving generic adviceB1
- Answer: The airport should reword the leading question and, at the plan stage, redesign the sampling method to draw future samples from a range of flights and times, not just one early-morning flight. At the collect stage, it should take steps to improve on the low 25% pilot response rate, for example by sending reminder emails or offering a paper option at the gate, and should contact passengers by more than one method so that those on other flights, or less likely to check email, are not under-represented in the main survey.
Question 58
- the claim is correctly judged to be false/not always trueB1
- a valid justification: beyond a certain size, a much bigger pilot reveals little extra information, since most wording or logistical problems would already show up in a smaller trialB1
- a different valid justification: a very large pilot uses up time, cost and potential respondents that were intended for the main survey, which works against the purpose of piloting cheaply firstB1
- Answer: The claim is false. Beyond a certain size, a much bigger pilot reveals little extra useful information, because most wording or logistical problems would already have shown up in a smaller trial. A very large pilot also uses up time, cost and potential respondents meant for the main survey, which defeats the point of piloting cheaply first.
Question 59
- (a) 3000 x 0.06, oeM1
- (a) 180 caoA1
- (a) Answer: 180
- (b) explains that 15 is too small: too few responses to reliably reveal problems or estimate a response rate, since results could easily happen by chanceB1
- (b) explains that 2500 is too large relative to the 3000 target: it would leave very few new residents for the main survey and would use up much of the time/cost that piloting is meant to saveB1
- (b) explains that 180 gives a reasonable balance: large enough to reveal wording/logistical problems and give a reasonably reliable response-rate estimate, while still leaving the great majority of the 3000 target sample available for the main surveyB1
- (b) Answer: A pilot of 15 is too small to reliably reveal problems or estimate a response rate, since results could easily be due to chance. A pilot of 2500 is too large, since it would leave very few new residents out of the 3000 target for the main survey and would waste much of the time and cost piloting is meant to save. A pilot of 180 strikes a reasonable balance: large enough to reveal genuine problems and give a fairly reliable response-rate estimate, while leaving most of the 3000 target sample for the main survey.
Mark your answers
Check answers without an account. Some answers can be checked automatically; use the mark scheme to mark the rest.