Exam 1 Study Guide
STAT 100
For each of the topics below, look over the relevant problems in your lecture notes and
homework.
EXPERIMENTAL DESIGN
Chapter 1 – Controlled Experiments
Main idea: Treatment and control groups should be as alike as possible.
• Randomized, double-blind design is ideal.
• Non-randomized and historical controls may introduce systematic bias.
Chapter 2 – Observational Studies
Main idea: Treatment and control groups are likely to be different in important ways since
they chose themselves.
• Difficult to conclude causation from association.
• Adjust for confounders by dividing into subgroups.
• Simpson’s Paradox: Clear-cut case of confounding where the overall percentages are
misleading and are reversed when divided into subgroups.
From a description of a study, you should be able to:
• Identify whether the study was a controlled experiment or an observational study.
• Identify what type of controls (if any) were used, i.e., randomized, non-randomized, or
historical.
• Determine whether the study was double-blind. Was a placebo used? Did the evaluators
know who was in the treatment and who was in the control group?
• Determine what conclusions are valid.
• Identify likely confounders and how you would adjust for them.
