single factor designs Flashcards
single factor - two levels
The simplest experimental designs have a single independent variable (or factor) with two levels of that variable. These designs can include between‐subjects factors or within‐subjects factors. Between‐subjects factors can be directly manipulated or they can be selected as subject factors. If manipulated, participants can be randomly assigned to groups (independent groups design) or matched on a potentially confounding variable, and then randomly assigned (matched groups design). If a subject variable is used, the between‐ subjects design is called an ex post facto design. Single‐ factor designs using a within‐subjects factor are usually called repeated‐measures designs (e.g., the famous Stroop studies). Studies using two levels of the independent variable are normally evaluated statistically with t‐tests (assuming interval or ratio data, normal distributions, and homogeneity of variance).
single factor - more than two levels
When only two levels of an experimental variable are compared, the results will always appear linear because a graph of the results will have only two points. Some relationships are nonlinear, however (e.g., the Yerkes‐Dodson law; the Ebbinghaus forgetting curve), and they can be discovered by adding more than two levels to an independent variable. Adding levels can also function as a way to test and perhaps rule out (falsify) alternative explanations of the main result. Like the two‐level case, multilevel designs can be either between‐ or within‐subjects designs.
analysing data from single-factor designs
Results can be presented visually in a bar graph when the independent variable is a discrete variable, or in a line graph if the variable is continuous. Studies using more than two levels of an independent variable are normally evaluated statistically with a one‐way analysis of variance or ANOVA (assuming interval or ratio data, normal distributions, and homogeneity of variance). A significant F ratio results in subsequent post hoc testing (e.g., Tukey’s HSD test) to identify precisely which means differ. Independent groups and ex post facto designs are evaluated with a one‐way ANOVA for independent groups; matched groups and repeated-measures designs are evaluated with a one‐way ANOVA for repeated measures.
special - purpose control group designs
In control group designs, the experimental treatment is absent for at least one condition. Varieties of control groups include placebo controls, often found in drug research; wait list controls, found in research on the effectiveness of a program or therapy; and yoked controls, in which the procedural experiences of the control group participants correspond exactly to those of the treatment group participants.
