L3 Surveys, sampling, weighted data Flashcards
what is a survey?
where data is (mainly produced) from asking standardised questions
who are the questions asked to?
a sample
what is a sample?
a fraction of a population under study
survey questions all use the same what?
wording
can surveys have open ended questions in them?
yes
what has to happen to these open ended questions later?
they must be coded
statisticians aim for samples which are ………… of the population
representative
self completion survey examples =
- postal survey
- mobile survey
- computer assisted web interviewing (CAWI / web)
- computer assisted self interview (CASI)
what does CAWI stand for?
computer assisted web interview
CASI =
computer assisted self interview
assisted / with interviews survey examples =
- pen + paper personal interview (PAPI)
- computer assisted telephone interviewing (CATI)
- computer assisted personal interviewing (CAPI)
PAPI =
pen and paper interview
CATI =
computer assisted telephone interviewing
CAPI =
computer assisted personal interviewing
why is using computers now the most common survey method?
to reduce human error
ethnographic study is a standardised measure which asks…
…everyone the same questions
what is a weakness of ethnographic studies?
time consuming + big commitment
how much does the census cost per person in the UK
£15
why do surveys?
as from the sample…we can estimate what it would be in terms of the whole population
what is the example of how we should imagine sampling?
like a soup
if the soup is stirred well (in terms of the sample) = a
random selection
then: a spoonful of soup will give a good idea of what the rest shall taste like = an
accurate result
when it comes to sampling; each member of the population should have an…
…equal chance of being chosen
what is the idea of sampling and the soup?
you don’t need everyone to determine what you’re trying to find out
in probability sampling: the selection relies on…
…chance
(probability sampling) the probability of selection =
known
(probability sampling) all member of the population have an …….. …….. of being included
equal chance
non probability sampling selection relies on…
…the researchers judgement
what might impact the researches sampling selection in non probability sampling?
- convenience
- practical reasons
in non probability sampling, the probability of selection =
not known
in non probability sampling, do all members have an equal chance to be included?
no
simple random sampling =
all equal chances of being selected
in simple random sampling, … ……… are considered
no criteria
stratified random sampling is where researchers want to…
…capture something from each subgroup
- i.e. ethnicities / age group
systematic random sampling is where =
sets of individuals are selected from a population at random
example of systematic random sampling =
i.e. picking a road and choosing every 15th house
(11, 26, 41, 56, 71, 86)
ATP =
American Trends Panel
EPSEM =
equal probability of selection method
CATI (computer assisted telephone interviewing) =
selecting a sample of area codes + randomly generating last 2 digits of telephone numbers
sampling error is where the sample does fails to…
…reflect the true population
when does sampling error occur?
when there is not enough variation in the sample
how can sampling error be controlled?
by the size of the sample
coverage error =
some members of the population are not covered by the sampling frame
what might cause coverage error?
when the sampling frame is poor / outdated
what does sampling bias depend on?
how well it represents the population
probability sampling CON =
- expensive
non probability sampling CONS =
- self selection
- overstudying of some panel members
true or false - all online surveys are created equally?
false
TP =
target population
weighting =
post work adjustments of survey data to make the sample more representative of the tp
weighting in the soup example =
seasoning the soup when its not quite perfect
what should the assigned weight never be larger than?
3
when can weighting only be applied?
if the probability of selection from the population is known
what is response error also known as?
‘non response bias’
response error is where -
people are less likely to fill in the survey
what happens if less people fill in a survey?
some opinions will be missed
why is response error becoming more of an issue?
as people do not want their data being collected
what makes a major impact on the quality of a survey?
the sample
convince sample is a …. ………… sampling method
non probability
- easier method for researchers to access a sample
