1 Experimental Design
2 Single Factor - One treatment with several levels. Multiple Factors - More than one treatment with several levels each. The Treatment is the independent variable (X) which we manipulate (change the level). The Response (which we observe/measure) is the dependent variable (Y).
3 Single Factor (One Way) Completely Randomized Design Between Subjects Design (Nested) Test objects (subjects) randomly assigned to only one level of the treatment. Easiest to design, understand, calculate, analyze. Relatively robust and free of restrictive statistical assumptions. Disadvantages - Requires large number of subjects Lacks sensitivity in detecting treatment effects
4 Single Factor - continued Randomized Complete Block Design Within Subject Design (Repeated Measures) Test objects (subjects) participate in all levels. Advantages - Require fewer subjects More sensitive in detecting differences Disadvantages - Adherence to strict statistical assumptions Subjects may undergo “learning effect”
5 Fixed Effects & Random Effects Fixed Effects - Treatment levels set to specific values. Results should only be generalized to these specific levels. Random Effects - Treatment levels randomly selected from a population of many different levels. Results may be generalized to the entire range of levels for that particular treatment. Note: A combination of fixed and random effects is sometimes referred to as a “mixed design”.
6 Multiple Factorial Design Two or more independent variables Factors = Treatments Each different factor (treatment) may have several different levels. Trial is an experiment at a specific level of a specific treatment.
7 Factorial Design Completely Randomized Factorial Design Between Subjects Design (Nested) Each subject participates in only one trial. Randomized Complete Block Factorial Design Within Subject Design (Repeated Measures) Each subject receives all levels of all treatments. Mixed Factor Design Between & Within Subject Design (Nested/Repeated) Subject participates in some, but not all levels.