Sampling
According to the research conducted by the Bureau of Justice Statistics (2015), an estimated 1.5 million prisoners were detained in federal or state correctional facilities across the US. Therefore, the large number of inmates held in prisons is the first challenge that needs to be dealt with regarding the research on mental and physical health needs of the incarcerated. Reducing bias during the process of sample selection will be possible through the implementation of random sampling that implies that each member of the population has an equal chance to be selected. However, because the population of inmates is scattered across the country, the pool of available subjects becomes skewed to the region where a researcher resides, thus causing bias in selection. On the other hand, random sampling can be used for selecting the correctional facility that will be involved in the research. For example, to account for both male and female inmates, a researcher should compose a list of all facilities and then differentiate them by type (male or female). Then, a random facility will be drawn from the lists of male and female prisons thus allowing a researcher to reduce the sample size to a much lower population number.
Because the number of prisoners in a correctional facility exceeds a few thousand, it is necessary to follow the selection of facilities with the selection of the actual research sample that will participate in the study of mental and physical health needs. Stratified sampling may be the most efficient and effective because it reduces error (Explorable, 2016). In this sampling method, a population is divided into different strata (groups). Then, a researcher randomly chooses the subjects for research proportionally from each population. It will be efficient to divide the population of inmates in prisons by age to account for the differences in mental and physical care needs. It is highly likely that the health care needs of an older population in prison will be much more diverse and specific than those of the younger population.
Stratified random sampling will allow the researcher to be more statistically precise in comparison with simple random sampling. This occurs because the variability within the strata is much lower than that of the entire population of prisoners. Furthermore, a stratified random sample cannot be considered biased because individuals chosen for research are selected randomly from the identified strata.
Research Design
The second scenario of conducting a study on the effectiveness of a new educational program for inmates and its impact on re-offending after release and the overall long-term educational achievement calls for finding former inmates that have already served time for their offenses. In this case, snowball sampling will be the most effective sampling method because potential subjects will be hard to locate. In snowball sampling, one participant will refer the researcher to the next participant and so on.
Regarding the research design type that will be appropriate for studying the chosen topic, it is possible that mixed research design – a combination of qualitative and quantitative – will be the most effective in addressing the threats of internal validity and selection bias. To collect data on the rates of recidivism, quantitative research can be applied while the long-term educational achievement can be studied through a qualitative method.
To observe a group of former inmates and their academic achievement, a cohort study can be implemented (a research program that investigates a group with a specific trait over a period) (Explorable, 2013). A retrospective cohort study will be the most appropriate – a researcher will analyze the historical data to judge the effects of a variable (an educational program for inmates).
References
Bureau of Justice Statistics. (2015). Prisoners in 2015. Web.
Explorable. (2013).Cohort study.Web.
Explorable. (2016).Stratified sampling. Web.