Sampling Flashcards
Usually we want to know about general/generalisable patterns/trends
Usually can’t collect data from everyone we’re interested in as it is
expensive
time-consuming
sometimes not actually possible
Ask a few people and hope they give a
good representation of ‘everyone’
Population:
group of people/organisations/whatever that we want to know about
Population may be a quite focused group, sometimes we talk about a TARGET POPULATION, e.g.
certain age range, people who drive cars, etc.
Sample is the
group of people/organisations/whatever from whom we collect data
N =
population size
n =
sample size
Complete enumeration =
everyone is counted
Sample distribution is not always a good representation of population distribution as not all data has been
collected
Want sample data to be
representative of the population we are studying
Sample data to be representative of population: everyone in the population has to have a chance of being selected into the sample.
This is called:
Probability sampling or random sampling
Sample data to be representative of population: if we systematically exclude anyone, then our sample will be
biased - not a good reflection of the population
Sample data to be representative of population: sampling variation is an inherent part research process. But this strategy should give us results that are, on the whole, a good indication of
what’s going on in the population
Probability sampling: every unit (case) in the population has a
known, non-zero probability of being selected into the sample
[PS]Simple random sampling: every unit has the
same probability of selection
[PS]Stratified sampling: selection probabilities set separately for
different strata (selections of the population)
[PS]Cluster sampling: sample larger groups that contain
a number of cases (e.g. sample whole schools and then study all children within a school)
[PS]Multistage sampling: combine a
number of different approaches
Non-probability sampling: some units (cases) in the population have unknown probabilities of being selected, and/or some have
no chance of being selected into the sample
[NPS] Purposive sampling: researcher uses his/her own judgement to
select subjects
[NPS] Convenience sampling: researcher uses some strategy that is
convenient, e.g. street polling
[NPS] Quota sampling: researcher finds a certain number (quota) of subjects with
key characteristics, e.g. so many males/females, so many of whatever age, occupation
Sample size: all else equal, larger samples give
more precise estimates of the things we want to measure
Sample size: smaller samples are more prone to being
skewed by unusual cases
we can use different units of analysis in our research - the choice is
important
we rely a great deal on gathering information from
samples to tell us about the populations to which they belong
different strategies for
sampling
beware bias in sampling: some cases which are part of the target population being
systematically excluded from the sample
All else equal, larger samples give us more
precise information