AP Statistics: Sampling and Experimental Design Notes

Fundamentals of Data Collection & Sampling

  • Population: The complete collection of all items or subjects of interest.
  • Sample: A subset of individuals or items selected from the population to be studied.
  • Census: Data collected from every individual or item in the entire population.
  • Observational Study: A study where investigators observe subjects and measure variables of interest without imposing treatments. Data can be collected by examining past records or by following subjects into the future.
  • Experiment: A study where researchers intentionally assign conditions, known as treatments, to experimental units to measure their effects on response variables.
  • Inference & Generalization:
    • Generalizations to a population are only appropriate when the sample is randomly selected or otherwise representative of that population.
    • Causation (cause-and-effect relationships) can only be established through a well-designed experiment with random assignment, never through an observational study.
  • Statistical Significance: Occurs when observed differences between treatment groups are too large to reasonably attribute to chance variation alone.

Sampling Methods

  • Simple Random Sample (SRS): A sampling method where every sample of size nn has an equal chance of being chosen from the population (e.g., using a random number generator to select ID numbers).
  • Stratified Random Sample: The population is divided into homogeneous groups (strata) based on a specific characteristic, and a simple random sample is taken independently from each stratum.
  • Cluster Random Sample: The population is divided into heterogeneous groups (clusters); whole clusters are selected at random, and all individuals within the selected clusters are surveyed.
  • Systematic Random Sample: Selecting every kthk^{\text{th}} individual from an ordered population after a random starting point.

Sources of Bias in Sampling

  • Bias: Systematic favoritism of certain responses or outcomes over others.
  • Undercoverage Bias: Occurs when a portion of the population has a reduced chance or no chance of being included in the sample.
  • Voluntary Response Bias: Occurs when a sample consists entirely of individuals who choose or volunteer to participate, overrepresenting strong opinions.
  • Non-Response Bias: Occurs when selected individuals cannot be reached or refuse to participate, and their responses would differ systematically from those who respond.
  • Response Bias: Occurs when errors or flaws in the survey instrument or environment (e.g., leading questions, untruthful responses) produce inaccurate data.
  • Convenience Bias: Occurs when a sample is chosen based on ease of access rather than random selection.

Principles & Designs of Experiments

  • Four Elements of Experimental Design:
    • Comparison: Comparing at least 22 treatment conditions (or a treatment against a control) to assess relative effectiveness.
    • Random Assignment: Using chance to assign treatments to experimental units to balance uncontrolled variables across groups.
    • Control: Keeping potential confounding variables constant across all treatment groups.
    • Replication: Assigning treatments to multiple experimental units to distinguish treatment differences from chance variation.
  • Experimental Designs:
    • Completely Randomized Design: Treatments are assigned to experimental units entirely at random.
    • Randomized Block Design: Experimental units are first grouped (blocked) based on a shared variable that could affect the response, then treatments are randomly assigned within each block.
    • Matched Pair Design: Experimental units are paired based on common characteristics (or each unit receives both treatments in random sequence), and treatments are randomly assigned within each pair.

Key Experimental Concepts & Variables

  • Explanatory Variable: The variable intentionally manipulated by researchers (whose levels or combinations of levels form the treatments).
  • Response Variable: The outcome variable measured after treatments are administered.
  • Experimental Units: The individuals, animals, or objects assigned to treatments.
  • Confounding Variable: A variable related to the explanatory variable that influences the response variable, creating a false association.
  • Control Group: A baseline group given no treatment or an inactive substance to measure baseline response.
  • Placebo & Placebo Effect: A placebo is a fake or inactive treatment. The placebo effect is a natural response occurring simply because subjects know they are receiving a treatment.
  • Blinding:
    • Single Blind: Subjects (or evaluators) do not know which treatment is being administered.
    • Double Blind: Neither the subjects nor the research personnel interacting with or assessing them know which treatment is being administered.