The static-group comparison design is a quasi-experimental design in which the outcome of interest is measured only once, after exposing a non-random group of participants to a treatment, and compared to a control group.
The objective is to evaluate the effect of this treatment (or intervention) which can be:
- a medical treatment
- a training program
- a policy change
- an environmental event, etc.
This design has 3 major characteristics:
- participants are NOT randomly assigned to either receive the intervention or not (i.e. it is a quasi-experiment)
- the treatment and control groups are measured at the same time
- NO measurements are taken before the intervention
The static-group comparison design is one step better than the one-group posttest only design which has 1 posttest measurement only and no control group (and thus is weakest of the quasi-experimental designs).
Next we will discuss the advantages, limitations, and provide an example where static-group comparison was useful in practice.
Advantages of the static-group comparison design
1. Advantages over randomized controlled trials (a.k.a. true experiments):
In some cases, randomization is not feasible because of:
- Ethical concerns: In cases where the intervention is: an exposure to a chemical or drug, an essential medical treatment, surgery, etc.
- Practical reasons: In cases when it is too late to randomize participants or collect pre-intervention measurements. But more generally, compared to a randomized experiment, a quasi-experiment is cheaper both in terms of time and money as it requires a smaller sample size and no follow-up of participants.
2. Advantages over other quasi-experiments:
Collecting only 1 post-intervention measurement provides some value in terms of the internal validity of the design:
- No instrumental bias: Because we are not taking multiple measurements over time, we won’t have to deal with errors due to the change in calibration of the instruments.
- Not susceptible to regression towards the mean: This is in contrast with other designs that require pre- and post-intervention measurements. Regression can happen in cases where participants are included in the study based on their extreme initial scores; The problem will be that their next measurement will naturally become less extreme, providing a fake effect that can be mistaken for the true effect of the intervention.
- No temporal bias: Since the outcome is measured after the intervention, we can be certain that it occurred after it, which is important for inferring a causal relationship between the two.
Limitations of the static-group comparison design
In general, static-group comparison is considered a weak design if the aim is to prove a causal association between the intervention and the outcome of interest. This is because the treatment and control groups have a high risk of being non-equivalent due to:
1. Selection bias
The lack of randomization means that the groups may be different at baseline, therefore the researcher may be comparing different groups to begin with. The problem is that participants may differ on some characteristics that may offer an alternative explanation of the outcome, therefore confounding the relationship between the intervention and the outcome.
2. Survival bias
This can be due to the differential rate of death or quitting between the study groups. This is different from selection bias in that the 2 groups were equivalent at baseline, but became different afterwards because those who received the intervention and those who did not have different drop-out rates.
Dealing with the limitations of the static-group comparison design:
Szafran, 2007 presented a way of dealing with these limitations by using poststratification weighting. In a nutshell, this method is based on statistical calculations to make the treatment and control groups more comparable with regards to background characteristics, therefore increasing the internal validity of the static-group comparison design.
Example of a study that used the static-group comparison design
In 2011, Friedberg and Sinderman wanted to study if the CDI, a tool to measure depression in outpatients, can also classify inpatient children as having depressive disorders or not.
So they brought data from the archive of a medical center, and selected children whose records contained a depression diagnosis at admission and a CDI score. Patients where divided into 2 groups:
- the treatment group: contains children diagnosed with any type of depressive spectrum at admission
- the control group: contains children diagnosed with any other psychiatric problem
The CDI scores of the 2 groups were compared and the study concluded that the CDI was in fact able to discriminate between the 2 groups.
3 remarks about this example:
- This is a retrospective study, so randomization was not an option.
- The intervention (a.k.a. treatment) was the depression diagnosis at admission, so the treatment was the disease!
- Collecting a pre-intervention measurement was impossible since the researchers had no access to any data before the admission.
- Krishnan P. A review of the non-equivalent control group post-test-only design. Nurse Res. 2019;26(2):37-40. doi:10.7748/nr.2018.e1582
- Campbell DT, Stanley J. Experimental and Quasi-Experimental Designs for Research. 1st Edition. Cengage Learning; 1963.