BIOS 7021 - Experimental Design Summary

Experiments address lurking variables and establish cause-and-effect relationships.

Experimental Units/Subjects: Individuals on whom the experiment is conducted.
Response Variable: Outcome measured (quantitative or categorical).
Factor: Explanatory variable, controlled by the experimenter.
Levels of a Factor: Values of the factor used to form different groups.
Treatment: Specific condition applied to units; either a single level or a combination of levels.
Experimental Design: Arrangement of treatments and assignment of subjects to groups.

Control Effects of Lurking Variables: Keep conditions uniform.
- Comparative experiments are essential to determine treatment effects.
- Control Group: Receives no treatment or a placebo for comparison.
- Placebo Effect: Improvement due to expectations, controlled via placebo.
- Double-Blind Studies: Neither subjects nor evaluators know treatment assignments.
Randomization: Randomly allocate subjects to treatments to prevent systematic bias.
- Use random number tables for assignment.
Replication: Apply each treatment to multiple subjects to reduce variability.

Completely Randomized Design: All subjects randomly assigned among all treatments.
Matched Pairs Design: Subjects paired by similarity; each pair receives different treatments.
Block Designs: Subjects grouped by a characteristic to control for that variable in treatment assignment.

Durability of Fabric: Tested across multiple cleansing agents and water temperatures using systematic treatment assignment.
Multiple Sclerosis Study: Controlled for gender by equal random assignment within gender blocks.

Institutional Review Board (IRB) oversight to prevent harm to subjects.
Informed consent required from all participants.
Confidentiality of individual data is crucial; only group summaries can be published.