Stats 240 Mid-Semester Test PART A 2021
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Stats 240 Mid-Semester Test
PART A: Survey Design and Conduct (Andrew Sporle)
2021
A1 A private research company has been commissioned by a political party to run a
poll on what the general public’s views about the quality of the local hospital services. Will they be required to get ethics approval for this survey? List the reasons for your answer (brief bullet points only)
NO
It is a survey of the general public (not patients) and doesn’t include health information.
It is a private company funded by a private institution so there’s no requirement from funder, research institution or employer
Lecture 6 slides 29-31
A2 The hospital mentioned in A1 serves a large and ethnically diverse community.
The political party who is commissioning the survey wants to know the opinions of the members of the public who are eligible to vote. What sampling frame could be useful for this target population?
Electoral roll
Lecture 3 slide 39
A3 The available sampling frame has address information for individuals. The
political party does not have a large budget for the research and wants the results within two weeks. What type of survey process would suit this situation? Give two limitations of this type of survey process in a study of this type. (brief bullet points only)
Any data gathering method that is quick and cheap (e.g. postal survey, online survey (just), telephone poll with random digit dialling).
Lecture 4 slides 15-21
A4 The researchers wanted to do a snowball sample using research interviewers. List
two advantages of using interviewers in the survey process.
Increased response rate, increased completion rate, reduces misunderstanding, can match language/culture, can explore answers, able to include more open ended questions - or anything else that sounds sensible
Lecture 4 and 5
A5 List the four (4) reasons why a good response rate is important for a survey.
Reduces MAJOR source of Error
Reduces source of BIAS
Efficiency ($ spent)
Client perception
Lecture 4 slide 6
A6 What type of missing data can be corrected by imputation using other data from
the same data source?
Missing at Random (MAR)
Lecture 4 slide 23
A7 What are the four (4) steps the researchers should take in evaluating the effect of
bias in a study?
Identify biases that may have influenced the result What would the likely direction of the bias be? How large might the effect be?
Include in your discussion of results.
Lecture 3 slide 11
A9 Which one of the following statements is false?
i. The more complex the survey process is, the more potential sources of non-sampling error there are for the survey
ii. Pre-testing a survey instrument should be done if the research team has developed their own questions
iii. The non-sampling error in a survey is often larger than the sampling error
iv. Increasing the sample size of a survey can result in making non-sampling errors larger
v. Official statistics surveys in New Zealand have high response rates because there is a legal requirement to complete them and StatsNZ invest a lot of money in good sample frames and participant recruitment processes.
vi. Post-stratification reweighting of a survey sample should be used to resolve issues of incompleteness in the sampling frame. (L3 S34)
vii. If the sampling frame population size is very much larger than the target population size, then the sampling process will require filter questions be asked of all those selected in the sample.
viii. The use of face-to-face interviews can be an excellent way to increase response rates and completion rates, but many survey don’t used them as they are so expensive and time consuming.
(Please circle one) (i) (ii) (iii) (iv) (v) (vi) (vii) (viii)
A10 Which one of the following statements is false?
i. Area-based measures of Socio-Economic Position (SEP) are often used as they are based on robust theoretical principles. (L5 S33-36)
ii. An individual can record multiple ethnic identities in New Zealand official statistics and these could differ across time and between Censuses
iii. A DAG is a causal pathway diagram with a single direction of causal flow.
iv. Occupation-based measures of Socio-Economic Position have a good theoretical foundation but are often not used as the occupation information has to be collected and coded.
v. The measure of socio-economic position that should be used in a study depends on the theoretical approach of the research and the timing of the association between the socio-economic position and the outcome.
vi. Comparing social outcomes between populations of different sizes needs to consider if the age structure of the populations is important to the outcomes being studied.
vii. Good questionnaire design can decrease bias by increasing completion rates
viii. If closed questions include an “other” response option, then the rest of the response options for the question should be designed to minimise the number of responses in the “other’ category.
(Please circle one) (i) (ii) (iii) (iv) (v) (vi) (vii) (viii)
A11 Which one of the following statements is false?
i. The requirement for ethical approval for research depends on who is doing the research.
ii. A survey can use a sampling frame from another study.
iii. The most recent Census can be a very useful source of information to test the representativeness of a sampling frame.
iv. A large proportion of New Zealand’s official statistics are produced to meet this country’s obligations to international agencies.
v. Official statistics administrative data can be cheaper to use than collecting data but requires a prior assessment of the suitability of the data definitions and collection process to the research question. vi. Official statistics administrative data is useful for research as it has no sampling error and no non-sampling error vii. The research approval process may require changes to your survey tool that can reduce the response rate viii. Statistical assessment is important in the ethical review of research as it can check that the study design can deliver statistically robust results and is as efficient as possible if there is any risk of harm to participants. (Please circle one) (i) (ii) (iii) (iv) (v) (vi) (vii) (viii) |
|
|
Total = |
17.5 |
|
2022-08-24