Rigor, purpose, and methodology in research design

¶ … Correctly Determining a Sample for Research

The prudent researcher must consider a plethora of factors when it comes to developing a sample for a research study. Although sample selection is one of the most important aspects of a research study, it occurs subsequent to a host of other processes that include a review of literature to determine a need for research, a hypothesis of what is believed to occur or to be proven within a study, as well as specific research design aspects that will demonstrate (ideally) the veracity of a hypothesis. All of these factors considerably influence how a researcher chooses to select his or her sample of the population to study. They play a substantial role in the determining of inclusion and exclusion criteria for the sample of a particular study. Other considerations include whether or not the study is qualitative or quantitative, as well as other inherent aspects of the study that will elucidate what specific type of sampling methodology the research will benefit from the most.

To that end, researchers should also base their development of a sample population subset on a number of threats that can compromise the results of the study they will perform. Although there are myriad types of threats that can seriously compromise the accuracy and validity of the data and their implications, the successful anticipation of these threats can allow researchers to facilitate any number of design features that can counteract the integrity of their work. The overarching point of virtually any research study is to establish causation for an effect in a manner that is most unambiguous as possible, since in several cases events can occur subsequent to one another and still not be causal (No author, no date, p. 3). Doing so requires the usage of controls so that there is a stark contrast between the independent and dependent variables that one is studying. Although there are quite a few types of controls that yield varying degrees of efficacy, research indicates that when relevant to a particular study, "randomization is the most effective method for controlling individual confounding variables" (Polit & Beck, 2008, p. 287).

Once study members have fostered a research design that should allow for the comparison of independent and dependent variables and taken an account of the varieties of threats that can compromise the integrity of the data collection process, it is then necessary to determine the most advantageous type of sample of a population to utilize for the study. Ultimately, what researchers are attempting to do when they determine a sample population is to avoid sampling biases that are the direct result of either an over-representation or an underrepresentation of the general population (Polit & Beck, 2008, p. 362). The goal of establishing a sample should be to have a group (of people, for most nursing studies) that is indicative of the general population that is affected by some aspect of the research design that the study is based on (Saldanha & Thompson, 2002, p. 257). Therefore, when selecting specific elements for a sample, researchers must take care to adhere to strict eligibility criteria in which they are sure to be able to accurately assess the different variables that the study's outcome hinges upon.

In establishing the inclusion or exclusion criteria that assists in the determination of a sample population, researchers must account for certain pragmatic concerns revolving around pecuniary issues, limitations of the population available to pool from, as well as other aspects of the research design that are pertinent to the study. Additional considerations include the specific resources that researchers have at their disposal to dedicate towards obtaining a sample of the population. By utilizing an eligibility criteria that systematically reduces the variables of the elements incorporated into a study, those conducting a study should be able to isolate and compare independent and dependent variables.

At this point in the research process, it becomes necessary to stratify the study into either a qualitative or a quantitative one. The subsequent type of sampling methodology employed will considerably vary depending on which type of study is conducted. Although it is possible to implement different types of study methods for different types of research studies, sample measures are either based on nonprobability sampling or probability sampling. If a researcher is performing a qualitative study, it is frequently advantageous to utilize one of the nonprobability sampling measures, which include different methodologies such as purposive, quota, or convenience sampling. However, when researchers are attempting to conduct quantitative studies, which usually utilize a deductive approach to research, (Creswell, 2002) it is oftentimes useful to employ some form of probability sampling, the likes of which include cluster sampling, simple random sampling, as well as systematic sampling.

The intrinsic virtue found in utilizing probability sampling with a quantitative study lies in the fact that most of these sampling measures utilize aspects of randomization which is extremely beneficial to researchers in that it provides a degree of control over variables that is unmatched by most other sampling measures. Additionally, it is highly noteworthy that probability sampling measures "yield more representative samples than nonprobability designs and permit estimates of the magnitude of sampling error" (Polit & Beck, 2008, p. 362). Depending on the particulars of the research design formed by a particular study, one of the most efficacious probability sampling measures in stratified random sampling, which enables researchers to categorize their sample into different types as well as induce essential elements of randomization that greatly increase the integrity of a study and minimize the threat of biases.

Although there is a definite possibility for incurring some form of bias when using nonprobability sampling methods, it is noteworthy that this potential for bias is somewhat reduced when one employs purposive sampling. This fact is largely due to the very nature of purposive sampling, which allows for researchers to individually select which members of a population that they would like to participate as elements in their sample. This sample methodology is most beneficial for qualitative researchers who are familiar with the population they are considering and have a fair amount of knowledge of the individual elements and the population as a whole. There are a host of different varieties of purposive sampling measures that can assist researchers in establishing causality and isolating independent and dependent variables, such as typical case sampling, maximum variation sampling, and extreme case sampling. The particular version of purposive sampling that works best for a specific research study depends on how much information researchers have about the populations, as well as on what type of research design they are employing and are attempting to test.