Intro to statistics
Lecture 2
Surveys
- Form of systematic and standardised data collection from a well defined population of interest
- Administered to a sample or entire population of people in order to describe and analyse attitudes, opinions,
behaviours or characteristics or population using statistical methods.
- Key characteristics of survey research;
- Sampling from population
- Collect data via questionnaires or interviews
- Design instruments for data collection
- Obtaining high response rate
- Generate quantitative data to be used in analysis
Population – the group of individuals having a characteristic that defines them from other groups
Sampling frame – actual list of sampling units from which sample is collected
- Occasionally get perfect match with target pop.
- More often imperfect – missing eligibles & ineligibles
- A source of ‘non sampling error’
Sample – group of participants in a study selected from population and agreed to take part
- Want as similar structure to pop. As possible
Probability sampling – selection of individuals from pop. So they are representative of population. Includes;
- Simple random sample
- All members of list have equal chance of getting picked
- Systematic sample
- Draw random sample by taking every nth case from list of elements
- Will be random if list is in random order
- Systematic random sample
- May be more convenient than simple random sampling
- Can introduce bias if pattern in how elements are listed in sampling frame
- Stratified sample
- Sampling frame is first stratified (divided into subgroups) before taking random sample from each group
(stratum)
- May be proportionate (random sample from each starta proportionate to its share of overall population) or
disproportionate (varies the sampling fraction between strata – may deliberately oversample from smaller
strata)
- Makes more likely sample is representative overall
- Cluster sample
- A compromise on randomised ideal
- Individual units of pop. Are often organised within hierarchies – use this structure – eg take random sample
of firms, then collect info from each worker in selected firm
Lecture 2
Surveys
- Form of systematic and standardised data collection from a well defined population of interest
- Administered to a sample or entire population of people in order to describe and analyse attitudes, opinions,
behaviours or characteristics or population using statistical methods.
- Key characteristics of survey research;
- Sampling from population
- Collect data via questionnaires or interviews
- Design instruments for data collection
- Obtaining high response rate
- Generate quantitative data to be used in analysis
Population – the group of individuals having a characteristic that defines them from other groups
Sampling frame – actual list of sampling units from which sample is collected
- Occasionally get perfect match with target pop.
- More often imperfect – missing eligibles & ineligibles
- A source of ‘non sampling error’
Sample – group of participants in a study selected from population and agreed to take part
- Want as similar structure to pop. As possible
Probability sampling – selection of individuals from pop. So they are representative of population. Includes;
- Simple random sample
- All members of list have equal chance of getting picked
- Systematic sample
- Draw random sample by taking every nth case from list of elements
- Will be random if list is in random order
- Systematic random sample
- May be more convenient than simple random sampling
- Can introduce bias if pattern in how elements are listed in sampling frame
- Stratified sample
- Sampling frame is first stratified (divided into subgroups) before taking random sample from each group
(stratum)
- May be proportionate (random sample from each starta proportionate to its share of overall population) or
disproportionate (varies the sampling fraction between strata – may deliberately oversample from smaller
strata)
- Makes more likely sample is representative overall
- Cluster sample
- A compromise on randomised ideal
- Individual units of pop. Are often organised within hierarchies – use this structure – eg take random sample
of firms, then collect info from each worker in selected firm
