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.