Lecture 6 - Sampling Flashcards
Purpose of Sampling
- Process of selecting observations
2 primary reasons:
* Not feasible to collect data from entire population
* Not necessary to collect data from entire population
2 types of Sampling
- Probability
- Each member of the target population has a known and equal chance of being selected into the sample (aka random selection)
- Reduces systematic error and bias
- Generalize to the greater population of interest
- Non-Probability
- The probability of each member of the target population being included in the sample is unknown
Key Components of a Sample
- Sample Element - Unit studied
- Population - Targeted group
- Sampling Frame - List of all elements in the population
- Population Paremeter - a value for a given variable in a population
- Sample Statistic - A value of a given variable in the sample
Example of Sample
Studying whether SFU students approve of constructing a gondola on campus
Target population: 37,000 SFU students
Sample Element: Student
Sampling Frame: list of 37,000 student numbers
Variable: Approval of Gondola - Scale from 1-5 (dissapprove to approve)
Random Sample: 100 students
True population average
Average of the sampling distribution = Average of an infinite # of sample averages ~= True population average
Estimating Sampling Error
Standard Deviation (SD):
* Spread of scores around the average in a single sample
Ex. 50 students chose 1, 50 students chose 5 vs. 100 students chose 3
Average = 3 in both but spread is different
Standard Error (SE):
* Standard deviation of the sampling distribution
Sample size increase = SE decreases
Random Assignment vs. Random Selection
A study can have one or the other, none, or both
Random Selection
* How you choose individuals from the population
* Related to sampling and generalizability
Random Assignment
* How you allocate chosen participants into different groups or conditions within study
* Related to study design and internal validity
Probability Sampling Methods
Assumptions:
* Defined population
* Sampling frame
* Pre-determined sample size
4 methods:
* Simple Random
* Systematic Random
* Stratified Random
* Cluster Random
Simple Random Sampling
- Foundation for unbiased sampling
Process:
* Establish sampling frame
* Each sample receives a number
* Determine intended sampling size
* Use random number generator/random number table to select which elements to be in the sample
Systematic Random Sampling
- More common for large samples
Process:
* Establish sampling frame and ensure list is in random order before numbering (protect against bias)
Determine sample size, calculate interval, and pick numbers according to calculated interval
Stratified Random Sampling
Ensures appropriate representation from identified subgroups
Process:
* Establish sampling frame, identify stratification variable and stratify population into subgroups
* Take simple random sample from each subgroup
Ex. If population is 100 - 48 men, 52 women then:
Stratified sample: Sample of 100 = 48 men, 52 women
Variation: Disproportionate
* Intentionally over or under sample specific subgroup to ensure representation
Cluster Random Sampling
Used when it is impossible or impractical to establish a sampling frame of individuals
Group population in clusters (e.g. geographic area)
Process:
* Establish clustered sampling frame
* Randomly sample from the list of clusters
* All elements within each selected cluster will then form your sample OR randomly select individuals from within each cluster
Sampling Frame
List or set of elements. It represents the “accessible” version of the targetted population
Ex.
Population = University students
Sampling Frame = Official enrolment list of students
Non-Probability Sampling Methods
Characteristics: Probability of selection is unknown
* No random selection
* Limited representativeness & generalizability
* Subject to researcher judgement
4 methods:
* Purposive Sampling
* Convenience Sampling
* Quota Sampling
* Snowball Sampling
Purposive Sampling
- Sample selection based on prior knowledge of subject or population (particular characteristics/attributes)
- Strategic selection of population elements
Ex. Specifically recruiting those: - In commited relationships
- Took a criminology class
Convenience Sampling
Relies on those who are readily available (Low cost and efficient)
* No designated population of interest
* Good approach if specific time and location is important to research question
Quota Sampling
Non-probability version of stratified random sampling
* Slect sample based on fixed quota for a particular characteristic/variable
Issues:
* Requires knowledge of the population which is difficult to accurately obtain (quota = proportionate to population)
* Biases may exist when selecting within each quota
Snowball Sampling
Identify 1+ participants who meet the eligibility requirements
* Rely on these participants to identify and refer others to participate
* Common for field observation or qualitative interview studies
* Not representative of population, but good for hard to reach populations
