Quasi-Experimental Design

Quasi-experimental studies have several issues that lead to lack of internal validity of the study. These occur as a result of the experimental conditions not being highly controlled or randomized. This leads to intervention and control groups being nonequivalent leading to issues with study design. This is what is evidenced in the study conducted by Yuan et al. (2009) where five units assigned a number of 8 to 10 voluntary participants to either the experimental or control groups. There is thought to be a risk of bias in assigning participants to the two groups as a result of this. This comes from the random error that occurs on measurements across the two different groups. This random error also affects the statistical conclusion validity

of the study since it creates an element of instability in the measurements making the results of the study to be highly unstable thus unreliable. Since there is a risk of bias with such quasi experimental designs, there is the introduction of a constant source of error into the results which creates the instability in the results.
In a quasi-experimental design there is also the risk of self-selection effect.

This occurs when participants choose the group which they want to join. In this study, it can be expected that those nurses who are more fit and willing to embrace physical fitness were more likely to opt for the intervention group rather than the control group. This creates an error since it becomes hard to estimate whether the effect of the intervention is as a result of the intervention itself or a preexisting factor in the participant which in this case is the likelihood of joining physical fitness programs (Moss, 2007).
Another threat to internal validity that may arise from this study by Yuan et al. (2009) is that of differential selection. This is where the study investigators are more likely to assign participants that fit a particular description to a particular group. For example it may be thought that those who were seen to be less fit were assigned to the intervention group in order to see if the intervention had any effect on this. This creates a variation in the groups that is much to the extreme (Sireci, 2007).
Strategies to strengthen the study’s internal

validity
The best strategy to strengthen the study’s internal validity is to have an experimental design where participants are randomly assigned into either the intervention or control groups.

In this case, the investigators would have to come up with some sort of random code with which the study participants are then randomized to either the intervention or control group. They should stick to this randomization code despite issues with self-selection or differential selection. Additionally after using a randomization code, the code should be adequately concealed in order to prevent the investigators and the participants from being able to tell what the group that the next participant will land. The investigators should also make efforts to blind the participants, investigators and those involved in assessing outcomes. This generally ensures that there is no way in which the effect estimate can be influenced by external factors. In this study it may be difficult to ensure blinding (Mislevy, 2007). However, it may be attempted by choosing nurses in different shifts or those in different facilities who may not have time to meet and discuss about the study. Blinding of outcome assessors such as those performing fitness evaluations in this study can be ensured by making sure they test the participants without knowing which group they are

in.
In order to improve the internal validity of the study, there is also need to design the control group as best as possible. This means that all factors that could affect the outcome of the study should be taken into consideration. For example, in this study, both the experimental and control groups were given the structured questionnaire to fill out. This way ensures that the two groups are equal in all possible ways apart from the actual intervention. In designing the control group, the investigators should have thought about having the control group perform a different activity such as eating or watching TV during the time when the intervention group is exercising. This may, however, be difficult to achieve considering the intervention was to exercise at home and the control group would also be required to perform the activity at home. Therefore there was no way to ensure this was done since it would invade privacy.
The factors that threaten internal validity also affect the other forms of validity either positively or negatively. Therefore, since these strategies will lead to internal validity of the study, it is expected that the other forms of validity, construct, external and statistical conclusion validity will also be ensured. The one issue that stands out regarding validity is that in ensuring internal validity by ensuring blinding, it may make the results not to be generalizable across situations. This occurs, for example, when the nurses from a private hospital and public hospital are considered or nurses from different town with different realisms such as food, health, diet, etc. However, this can be ensured through including participants from as many regions as possible and ensuring randomization (Embretson, 2007).