Within classroom design

In a within classroom design, students are randomly assigned to conditions. From a strictly experimental viewpoint, this design is preferable since differences found are either due to statistical chance or due to experimental variation. In contrast, in a between classroom design, differences between conditions can also occur since conditions already differed prior to instruction.

However, in case of intrusive conditions where differences between conditions are quite obvious, a within classroom design can also reduce the study's internal validity (see Cook & Shadish, 1994). Given these circumstances, a between classroom design might be preferable. Possible threats to internal validity caused by participants' knowledge about the different conditions are:

• resentful demoralization: Students of a control condition might get demotivated when realizing that they are in the less preferable condition. This yields reduced efforts and thus an overestimation of the treatment effect.
• compensatory rivalry: Students of a control condition might try to make up for the disadvantage by putting extra efforts into their work. This results in an underestimation of the treatment effect.
• treatment diffusion: When students of a control condition get to know about the treatment in the experimental condition, they might imitate the behavior students of the experimental condition are supposed to show (e.g. reducing their gaming behavior, paying extra attention to giving elaborated feedback, ...). This will reduce the treatment effect.

Cook, T. D., & Shadish, T. R. (1994). Social experiments: Some developments over the past 15 years. Annual Review of Psychology, 40, 545-580.