Principles of Experimental Design
Apply the three principles: control (keep all other variables constant), randomization (eliminate confounding), and replication (enough experimental units for reliable results). Design completely randomized, block, and matched-pairs experiments.
A lesson of AP Statistics, in Collecting Data, about 10 minutes.
Experimental Design
By the end of this lesson, you will be able to identify the key principles of a well-designed experiment, design a completely randomized experiment or a block design, and explain the role of control, randomization, and replication.
Key Principles of Experiments
1. Control: hold all variables constant except the treatment. Establish a control group (receives a placebo or standard treatment) as a baseline for comparison.
2. Randomization: randomly assign subjects to treatment groups. This distributes known and unknown confounders equally across groups.
3. Replication: use enough subjects in each treatment group that chance variation is reduced and real effects can be detected.
Experimental Vocabulary
Experimental units (subjects): the individuals to whom treatments are applied.
Factor: an explanatory variable that is manipulated by the experimenter.
Level: a specific value of a factor.
Treatment: a specific combination of factor levels applied to experimental units.
Response variable: the outcome that is measured.
Control group: receives a placebo (or no treatment) to provide a baseline.
Placebo: an inactive treatment that looks like the real treatment.
Placebo effect: subjects improve simply because they believe they are receiving treatment.
Blinding
Single-blind: subjects do not know which treatment they received. Reduces placebo effect.
Double-blind: neither subjects nor those evaluating the response know which treatment each subject received. Eliminates both placebo effect and evaluator bias.
Completely Randomized Design
Subjects are randomly assigned to treatments with no restrictions. The simplest design.
Example: 60 patients randomly assigned to drug A, drug B, or placebo (20 each).
Randomized Block Design
Block: a group of experimental units that are similar in some way that may affect the response. Subjects are randomized within blocks, not across the entire study.