Basic Concepts Flashcards
What is a sample?
a proportion of all the accessible information on members of the population
What’s an element?
smallest individual parts or items that make up the target population or sample.
What is the target population?
the entire collection of the elements about which we wish to make inference on.
What is the survey population?
the population we actually sample from; the population we realistically have access to.
> can be the same as target population in ideal scenario
What is a population parameter?
a mathematical function of the values of the variables across the whole population.
What is the sampling frame?
a complete and up-to-date list of all the accessible elements in the population
> what we sample from
> usually a subset of the survey population
what is a sample statistic?
a mathematical function of the variables observed in the sample.
What are sampling units?
the minimum unit of observation for information on the operative variables, i.e. people or households.
What are some aspects to avoid in questionnaires/surveys?
1) technical jargon
2) ambiguous questions
3) combined questions
4) leading or loaded questions
5) double negatives
What are the advantages of open response questions?
a) variety of possible responses
b) responses obtained with no influence
c) ideal for pilot study - gain understanding of possible responses
What are the disadvantages of open response questions?
a) variability in clarity and depth of responses
b) time consuming
c) difficult to code responses (thematic analyses required instead)
d) responses not directly comparable
What are the advantages of closed response questions?
a) easier to answer - reduces refusals
b) easier to present results
c) responses directly comparable
What are the disadvantages of closed response questions?
a) provides fewer opportunities for self-expression
b) survey designers may disagree on included responses
What must be considered when designing multiple choice questions?
a) categories must be exhaustive and mutually exclusive
b) include don’t know or NA option
How should you analyse likert scale data?
As if it were categorical unless its paired
How should questions be laid out?
1) start with straightforward questions
2) questions that require thought/cover sensitive topics should be kept until the end
What is sampling error?
the variation/error in the result associated with examining a sample, not a survey of the whole population.
How do you quantify sampling error?
a) standard error of associated estimates
b) only for random samples
What is non-sampling error?
a) This is error due to the result of factors other than taking a sample, i.e. due to a flaw at some stage in the survey process.
b) extent cannot be measured so must be minimised in survey design
What are the two types of non-response error?
1) non-participation
2) missing/incomplete answers
How do you control non-response error?
1) suitable method of data collection
2) careful question design
3) skilled interviewers
4) incentives
5) reminders
What causes inaccurate response/response errors?
1) errors of definition
2) asking sensitive questions
3) asking leading questions
4) recording/processing errors
What is coverage bias?
a) when the sampling frame is too small and unable to represent the target population
What are the advantages of face-to-face interviews?
a) interviewer can explain and clarify
b) data can be entered directly and checked in situ
c) high response rate
