Given Hapunda

95 criteria (for example, matching programme non-participants with participants, individually or by aggregate). Without random assignment, research participants do not have the same chance of being assigned to the intervention or control group. Thus, it may be difficult to clearly establish the causal relationship between the intervention condition and observed outcomes. This is because there could be other factors or explanations (such as extraneous or confounding variables that cannot be controlled or accounted for) than the intervention programme. Although quasi-experimental designs eliminate more of the threats to internal and external validity than pre-experimental designs, they are not as strong as experimental designs in establishing causality. They have limitations with regards to internal and external validity (Marlow, 2011; Morra-Imas & Rist, 2009; Rossi et al ., 2004; Rubin & Babbie, 2013). Quasi-experimental studies are often more realistic in service delivery settings (Project Star, 2006). For example, a group of children who receive tutoring can be compared to the other children in their class who did not receive tutoring to see if grades are generally better in the tutored group. However, because randomisation is not used, one validity threat to quasi-experimental studies is “selection bias.” It may be that there is something different about the people who choose to participate in the programme (for example, they are more motivated) that makes them more likely to succeed. How can we be sure that the change we see (improved grades) was caused by the service and not this personal characteristic? Advantages of Quasi-Experimental Designs Quasi-experimental designs are more feasible or practical than randomised experiments because pure experimental designs are not easy to set up and execute as they require random assignment of subjects to groups (experimental and control groups), but quasi-experimental designs do not require randomisation. The quasi-experimental designs only require the creation of comparison groups which are not equivalent (Creswell, 2014; Morra-Imas & Rist, 2009; Marlow, 2011; Rubin and Babbie, 2013). More so, quasi-experimental designs can be conducted in the natural setting or environment of which results from the study can be applied to other settings and populations. This can allow generalisation of the findings from the sample to the general population. Quasi-experiment studies also allow the researcher to make any manipulations he or she desires, and thus the experimenter has control over the manipulations as they do not occur on their own. More so, using quasi-experiments allows experimenters to take into consideration ethical issues when carrying out the study (DeRue, et al ., 2011).