Between classroom design

From Pslc
Revision as of 16:13, 30 October 2007 by Koedinger (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

In a between classroom design, classes are randomly assigned to conditions. From a strictly experimental viewpoint, this design is rather detrimental: In comparison to a within classroom design, internal validity is reduced since differences can occur not only due to experimental variation and statistical chance, but also due to prior group differences (see Cook & Shadish, 1994). Therefore, some scientists argue that data of between design studies should be analysed at the group level, yielding smaller sample sizes and reduced statistical power.

However, in some situations, a between design is the best choice to make. Given rather intrusive study conditions in a within classroom design, the chance is high that participants will get to know about the other conditions' treatment. This can affect the study's validity. Thus in this case, a between design is preferable. In order to control for possible prior group differences, conditions should be compared with regard to descriptive variables that influence the outcome (e.g. prior knowledge, gender, age, ...). If differences are found, these should be controlled statistically (covariate analysis, see e.g. Huitema, 1980).


Cook, T. D., & Shadish, T. R. (1994). Social experiments: Some developments over the past 15 years. 'Annual Review of Psychology, 40,' 545-580. Huitema, B.E. (1980). 'The analysis of covariance and alternatives.' New York: Wiley and Sons.