SAMPLING DESIGNS Flashcards
population
entire aggregation/groups of people that meet a set of criteria
accessible population
aggregation meet criteria and people are actually accessible
target population
aggregate cases about which you want to make generalizations
sampling
process of selecting a portion of a population to represent an entire population
sample
actual subset of units that compose the population
representativeness
key characteristics of your sample are the same as the population
strata
mutually exclusive segments of population established by 1 or more characteristics
sampling bias
systematic over representation or under representation of some segment of the population with respect to a characteristic that’s relevant to the research
probability sampling
- make sure people in sample have equal chance of being picked to be in study
- some form of random selection in choosing the elements
- researcher is in a position to specify probability that each element of population will be included in sample.
non-probability sampling
- elements are selected by nonrandom methods
- no way to estimate probability that each element has of being included and every element usually does not have a chance for inclusion
non-probability sampling methods
NOT RANDOM
- convenience sampling (snowball/network)
- quota sampling
- purposive or judgmental sampling
- theoretical sampling
convenience sampling
don’t have access to people so you use what is available
quota sampling
research identifies strata of population and determines portion of elements needed from various segments
snowball/network sampling
- someone know someone that knows someone
- building
purposive/ judgmental sampling
based on belief that researcher knowledge of population can be used to hand pick cases that are to be included in sample
theoretical sampling
pick people that we know use instrument
probability sampling method
- simple random sampling (stratified rand
- cluster sampling
- systematic sampling
simple random sampling
researcher establishes sampling frame
sampling frame
actual list of elements from which sample will be chosen
stratified random sampling
use strata characteristics and make sure you have equal amount of people from each group
proportionate stratified sampling
size of sample strata is proportional to the size of population strata
disproportionate sampling
if there are not enough people, use a portion to represent the large population
cluster sampling
typically send in a large scale of surveys when all other methods become expensive
systematic sampling
selection of every k case from some list/group
can be both non/probability
sampling error
difference between population values and sample values
use power analysis
statistical procedure
need info on alpha
risk you want to take on a type I error (wrongly rejecting true null hypotheses) usually 0.05
need info on 1-beta which is standardly 0.80 = power
willing to take a 20% chance of comminting a type II error (wrongly accepting false hypothesis)
need info on gamma = effect size =
estimate of magnitude of relationship between research variables
if relationship between independent and dependent variables is strong, then need small sample
if weak need larger sample
use previous research to estimate
effect size
if no effect size then
- 20= small effect size
- 50= medium effect size
- 80= large effects size
attrition
drop out rates/dying
murphy’s law
if anything can go wrong, it will
of variables
larger # of variables, larger sample size, more # you need
sensitivity of measures
different measures vary in their ability to measure precisely concepts under study. if measure is vague, larger sample.
subgroup analysis
if sample is divided to test for effects in specific group
-sample must be large
steps in drawing a sample
- identify target population
- identify accessible population
- specify eligibility criteria (specific characteristics)
- specifiy sampling plan
- recruit sample
