Observational Studies Versus Experiments

Learn observational comparisons, random assignment.

Lesson 3 of 32 parts Practice Flashcards

1/2

Observational Comparisons

Objective: Explain observational comparisons, apply the stated method, and check its conditions.

Observational studies measure existing exposures without assigning them. A group-mean difference describes association and can be influenced by confounding or selection. Careful design and adjustment can help, but an unadjusted comparison alone is not a randomized causal effect. Association alone does not establish causation.

Formulas

Observed difference = group mean A − group mean B

Worked examples

Try itIn a synthetic observational study, Group A’s mean is 13 and Group B’s is 7. Find the observed difference.Show answer

Step 1: Subtract group means to describe an association. Step 2: Compute (13)−7. Step 3: The difference is 6; without a suitable design this does not establish causation.

Swipe sideways to move between sections

Observational Comparisons

Objective: Explain observational comparisons, apply the stated method, and check its conditions.

Formulas

Observed difference = group mean A − group mean B

Worked examples

Try itIn a synthetic observational study, Group A’s mean is 13 and Group B’s is 7. Find the observed difference.Show answer

Step 1: Subtract group means to describe an association. Step 2: Compute (13)−7. Step 3: The difference is 6; without a suitable design this does not establish causation.

Random Assignment

Objective: Explain random assignment, apply the stated method, and check its conditions.

Random assignment allocates experimental units to treatment conditions and supports causal comparison by balancing confounders in expectation. It differs from random sampling, which supports population generalization. A volunteer experiment may be causal for its participants without automatically representing everyone. Random assignment and random sampling serve different purposes.

Formulas

Random assignment supports causal comparison; random sampling supports generalization

Worked examples

Try itA synthetic randomized experiment assigns 14 volunteers equally to treatment and control. Find each group’s size.Show answer

Step 1: Random assignment allocates experimental units to conditions. Step 2: Divide 14 by 2. Step 3: Each group has 7; random assignment alone does not make volunteers representative of everyone.

You finished the reading!

Lock it in with a few practice questions. This is the last lesson in the chapter.

Practice

Previous lessonBias

Last lesson