What is a survey?
- Target groups
- We are interested in target population
- We’re not interested in individuals, instead groups of individuals and the characteristics of these groups
- Sample:
- A selection of units from the target population.
- We collect data from the sample of people.
- We estimate the result. We don’t do a census, but we estimate it.
- We are using statistics for it:
- Standart error, variance
- Confidence interval
- Sample should be random sample.
- Composition of the sample should be similar to the composition of the target population.
- Everyone from the target population has the same possibility to be selected.
- Techniques of sampling is important.
Total Survey Error
- Measurement: From theoretical concept to the respondent’s answers
- construct
- come up with a definition of our “question” → which is in this case “smoking”
- limit ourselves to the key components of “smoking”
- validity error
- operationalization: translate the terms we are using in our questionarre to
- measurement
- response
- while trying to correct measurement errors, we might create “processing error”
- edited response
- Representation: How the group of respondents is selected and weighted.
- target population
- usually not everyone is included, there are exceptions
- when making these adjustments (who to include, who to exclude) we might have error
- there is no right or wrong, we should only document of what we have done
- for instance “a clear definition of our target population”
- sampling frame
- is often incomplete
- how to invite our sample to the survey?
- they might not have mobile phones, then we are excluding a part of the population 20%
- coverage error: describes the completeness of the sampling frame
- percentage itself is not the problem, but it is how they differ from the rest of the population
- a person without a mobile phone might differ drastically
- sample
- non-response error:
- refers to incompleteness of the sample
- non-contact error is a component
- respondents
- adjustment error:
- we have to compensate the non-response error
- weighting is used to compensate
- statistical quality shrinks
- effective sample size shrinks: statistical power of sample
- postsurvey adjustments
- Variance (coverage, meausrement)
- Random error
- Precision of estimate
- Reliability
- No group is underrepresented
- Bias
- Systematic error
- Accuracy of estimate
- Validity
- 4 Key components of survey quality:
- Measurement Error
- Sampling Error
- Nonresponse Error
- Coverage Error