Surveys Flashcards
What are surveys? What sort of dataset are they most useful in gathering?
Surveys are a method of gathering data from LARGE SAMPLES through responses to a structured set of questions. These tend to be standardised.
What can surveys be useful for?
Surveys can be useful for description and for explanation. They are used for observational research.
What do surveys of sample populations aim to produce? Why is this often difficult?
Surveys of sample populations aim to allow general claims to be made about a whole population or a sub-group. This is often made more challenging by sampling biases (who is selected for surveys) or by measurement error (what is measured by the information given)
What is probability sampling? How does this relate to the law of large numbers?
Probability sampling is a sampling method whereby a whole population has a non-zero chance of being selected in a sample. This works with the law of large numbers, where with random selection, a sample mean will increasingly approximate the population mean as the sample size increases.
What is simple random sampling? What issues are there with this?
Simple random sampling is where respondents are randomly selected from a larger group. This does not show sub-groups especially well, however.
What is stratified sampling? What issue is there with this?
Stratified sampling is the process whereby a population is partitioned into subpopulations with notable distinctions. These units are then randomly sampled from different strata.
This may overrepresent particular groups, damaging its generalisability.
What is systematic sampling?
Systematic sampling is the process whereby participants are selected at regular intervals from a population, following pre-determined rules.
What is non-probability sampling? What types of this are there and what issues are there with this?
Non-probability sampling is a method used where participants are hand-picked rather than randomly selected. This includes quota sampling, volunteer sampling and snowball sampling. Issues arise in that it is more difficult to form a population inference, and the probability of selection of units into the sample is unknown.
What are nominal, ordinal and quantitative/continuous questions respectively?
Nominal - binary, yes/no, or multiple choice questions
Ordinal - scale questions, behavioural frequency
Quantitative/continuous - scale from 1-10
