The static group comparison

Another simple pre-experimental design is the static group comparison, which can be represented as follows:

This is a simple comparison between an experimental group which receives a treatment (X) and a control group with no treatment, with O1 and O2 indicating measurements on the experimental units taken after the treatment has been applied. Here, the basic problem is that there is no way of knowing whether the two groups were comparable before the treatment. Therefore, treatment effects that are measured by the mean value for treated units minus the mean value for control units may really be differences that would have been there anyway.

Since there was no "before" (pre) testing, we really don't know much about the groups prior to sampling:

  • Although randomly selected, there is the chance that their behavior after exposure to the stimulus is quite consistent with how they were acting prior to exposure
  • In other words, we may infer causes based on group difference(s) that are actually NOT the result of exposure to the critical event;
  • Such differences are possibly due to characteristics that existed in the population prior to testing

For a static group comparison with X denoting attrition, we may take the above points as unilateral null axioms: Consistency of behavior prior to attrition, differences are not result of attrition, such differences existed prior to testing. Again, such difference would have to be measured external to the conditions of attrition. Formalization of static group w/ attrition remains unchanged, unlike the one group pretest-negative-posttest design via attrition ‘treatment’. 
... ? ?