211-213
Sampling and Confidence Intervals
Random Sampling and Population Estimates
A random sample of 100 registered voters leads to estimating that 50% approve of a proposed assault rifle ban.
Confidence Level: The probability that the true value lies within a specified range (confidence interval).
Example Confidence Interval: 40 - 60% (95% confidence), 45 - 55% (68% confidence).
Understanding Sampling Error
Sampling error calculation informs us about the reliability of our sample estimate.
Population size is often irrelevant; accuracy depends on how the sample is chosen, not its size relative to the population.
Cautions in Probability Sampling
Assumptions of probability theory may not hold in real-world surveys due to factors like non-response and sampling techniques.
Researchers should be cautious in generalizing results beyond the sampling frame, as samples may not represent the entire population.
Probability Sampling Designs
Types of Probability Sampling
Various designs for different research purposes (simple random, stratified, cluster, systematic sampling).
Effectiveness of Membership Lists
Membership lists (like organizations) can provide a complete sample frame if all members are represented.
Researchers should understand potential incomplete or biased listings and adjust generalizations accordingly.
Sampling Frames and Practical Considerations
Defining Sampling Frames
A sampling frame is the actual or theoretical list from which a sample is drawn, often essential for accurately studying a defined population.
Examples include licensed drivers or membership lists which provide ideal contexts for research.
Limitations of Telephone Directories
Directories can miss new subscribers, exclude unlisted numbers, and contain non-residential listings.
Approximately 48% of adults live in households with only wireless phone service, complicating sample accuracy.
Practical Examples of Sampling
Applications in Research
Example: Using the American Correctional Association membership for studying corrections administrators.
Operational Definitions
Sampling frames act as real-world definitions of the abstract population being studied, crucial for accurate results.
Simple Random and Systematic Sampling
Simple Random Sampling
Basis of probability theory; each member of the population has an equal chance of selection.
Utilizes random number tables or computer programs for selection efficiency.
Systematic Sampling
Selects every nth element from a structured list, enhancing efficiency.
Care must be taken to avoid periodicity in lists that could bias samples based on the arrangement of the data.
Potential Bias in Systematic Sampling
If elements are cyclically arranged (e.g., premium apartments on specific floors), bias can occur if the sampling interval matches the periodicity.