Survey design and sampling methods

Likert scale

type of rating scale that measures respondant’s level of agreement with a statement

BRUSO

criteria for good survey question writing - brief, relevant, unambiguous, specific, and objective

common errors with survey questions

leading questions, double-barreled questions, negatively worded questions, question order

compare probability and nonprobability sampling

In probability sample,all members of a population have an equal chance of getting selected in the sample (usually using a sampling frame). In nonprobability sample, there is no way to know the probability of any one person getting included in the sample.

types of probability sampling methods

simple random sample; stratified random sample; cluster sampling

compare disproportionate and proportionate stratified sampling

proportionate stratified sample tries to ensure proportions of strata in sample match those in the population - this isn’t the case in disproportionate sample (might priotize having equal numbers from each strata, even if population is not equal).

sampling frame

a list of all members of the population of interest

cluster sampling

population is seperated into homogenous “clusters” and clusters are randomly choosen to be included in sample.

types of nonprobability sampling

convenience sampling, self-selecting sample, purposive sampling, snowball sampling,, quota sampling

purposive sampling

nonprobab sampling methods - researchers predetermine that they want to study a specific group and then sample from an area that those people would frequent.

snowball sampling

a type of urposive sampling where one participant recommends another

quota sampling

research tries to get a specific number of participants in each category using nonprob methods (a variation of stratified sampling - but not random)

sampling bias

sample is not representative of the population (results are not accurate) - often due to nonprobability sampling

nonresponse bias

non-responders differ systematically from responders

what are shortcuts used in survey response?

response sets like acquiescence (yea-saying), fencing-sitting, desirable responding (faking good), faking bad

acquiescence

yea-saying - answering yes/strongly agree to all survey questions - check using “reverse wording” questions

fence-sitting

answering with the middle option/neutral option for all question - can remove middle option, but this leaves out respondants who really are neutral.