Controlled Experiments and Causation

Experiments vs. Observational Studies

Response Variable

Sample Survey

Census

Randomization Increases the Scope of Inference by Generalizing to the a Larger Population

Identify Sample and/or Population

Experimental Units

Assigning a Number to Each Participant to Randomly Select

Randomization

Replacement vs. Non-Replacement

Simple Random Sampling (SRS)

Stratified Sampling

Cluster Sampling

Undercoverage Bias or Measurement

Sampling Variability and Bias

Sampling with a Diagram

Choosing the Most Appropriate Sampling Method

Potential Confounding Variables, Or is it?

Non-Response Bias

Response Bias

Self-Selection Bias

Convenience and Voluntary Response Bias

Selection Bias

Sampling Bias and Errors

Biased Estimator

Unbiased Estimator

Control to Show a Change by Treatments

Replicate to Show That Your Results Didn't Happen by Chance

Blind and Double-Blind to Reduce Bias

Experimental Design and Control for Confounding Variables

Only Well-Designed Experiments Show Cause-&-Effect

Blocking for Homogenous Groups to Reduce Variability

Treatments

Use Placebo to Show Placebo Effect and to Reduce Bias

Treatments to Explanatory Variables

Factors & Treatment Levels

Completely Randomized Experimental Design

Randomized Block Design

Matched Pairs and Other Experimental Designs

Experimental Flaws Reducing Statistical Significance

Statistical Significance