Sampling

CHAPTER 12


Sampling


Judith Haber




image


Go to Evolve at http://evolve.elsevier.com/LoBiondo/ for review questions, critiquing exercises, and additional research articles for practice in reviewing and critiquing.


The sampling section of a study is usually found in the “Methods” section of a research article. It is important for you to understand the sampling process and the ingredients that contribute to a researcher using the most appropriate sampling strategy for the type of research being conducted. Equally important, is knowing how to critically appraise the sampling section of a study to identify how the strengths and weaknesses of the sampling process contributed to the overall strength and quality of evidence provided by the findings of a study.


When you are critically appraising the sampling section of a study, the threats to internal and external validity as sources of bias need to be considered (see Chapter 8). Your evaluation of the sampling section of a study is very important in your overall critical appraisal of a study’s findings and their applicability to practice.


Sampling is the process of selecting representative units of a population in a study. Although sampling is a complex process, it is a familiar one. In our daily lives, we gather knowledge, make decisions, and formulate predictions based on sampling procedures. For example, nursing students may make generalizations about the overall quality of nursing professors as a result of their exposure to a sample of nursing professors during their undergraduate programs. Patients may make generalizations about a hospital’s food or quality of nursing care during a 3-day hospital stay. You can see how exposure to a limited portion of these experiences forms the basis of our conclusions, and how much of our knowledge and decisions are based on our experience with samples.


Researchers also derive knowledge from samples. Many problems in research cannot be solved without employing rigorous sampling procedures. For example, when testing the effectiveness of a medication for patients with asthma, the drug is administered to a sample of the population for whom the drug is potentially appropriate. The researcher must come to some conclusions without giving the drug to every patient with asthma or laboratory animal. But because human lives are at stake, the researcher cannot afford to arrive casually at conclusions that are based on the first dozen patients available for study.


The impact of arriving at conclusions that are not accurate or making generalizations from a small nonrepresentative sample is much more severe in research than in everyday life. Essentially, researchers sample representative segments of the population because it is rarely feasible or necessary to sample the entire population of interest to obtain relevant information.


This chapter will familiarize you with the basic concepts of sampling as they primarily pertain to the principles of quantitative research design, nonprobability and probability sampling, sample size, and the related appraisal process. Sampling issues that relate to qualitative research designs are discussed in Chapters 5, 6, and 7.




Sampling concepts


Population


A population is a well-defined set that has specified properties. A population can be composed of people, animals, objects, or events. Examples of clinical populations might be all of the female patients older than 65 years admitted to a certain hospital for congestive heart failure (CHF) during the year 2013, all of the children with asthma in the state of New York, or all of the men and women with a diagnosis of clinical depression in the United States. These examples illustrate that a population may be broadly defined and potentially involve millions of people or narrowly specified to include only several hundred people.


The population criteria establish the target population; that is, the entire set of cases about which the researcher would like to make generalizations. A target population might include all undergraduate nursing students enrolled in accelerated baccalaureate programs in the United States. Because of time, money, and personnel, however, it is often not feasible to pursue a study using a target population.


An accessible population, one that meets the target population criteria and that is available, is used instead. For example, an accessible population might include all full-time accelerated baccalaureate students attending school in Ohio. Pragmatic factors must also be considered when identifying a potential population of interest.


It is important to know that a population is not restricted to humans. It may consist of hospital records; blood, urine, or other specimens taken from patients at a clinic; historical documents; or laboratory animals. For example, a population might consist of all the HgbA1C blood test specimens collected from patients in the Upper City Hospital diabetes clinic or all of the patient charts on file who had been screened during pregnancy for HIV infection. It is apparent that a population can be defined in a variety of ways. The important point to remember is that the basic unit of the population must be clearly defined because the generalizability of the findings will be a function of the population criteria.



Inclusion and exclusion criteria


When reading a research report, you should consider whether the researcher has identified the population characteristics that form the basis for the inclusion (eligibility) or exclusion (delimitations) criteria used to select the sample—whether people, objects, or events. The terms inclusion or eligibility criteria and exclusion criteria or delimitations are applied to define attributes that restrict the population to a homogenous group of subjects. The population characteristics that provide the basis for inclusion (eligibility) criteria should be evident in the sample; that is, the characteristics of the population and the sample should be congruent. Examples of inclusion or eligibility criteria and exclusion criteria or delimitations include the following: gender, age, marital status, socioeconomic status, religion, ethnicity, level of education, age of children, health status, and diagnosis. The degree of congruence between the criteria and the population is evaluated to assess the representativeness of the sample.


Think about the concept of inclusion or eligibility criteria applied to a study where the subjects are patients. For example, participants in a study investigating the effectiveness of a motivational-interviewing–based coaching intervention compared to usual care to improve cancer pain management (Thomas et al., 2012; see Appendix A) had to meet the following inclusion (eligibility) criteria:



Remember that inclusion and exclusion criteria are established to control for extraneous variability or bias that would limit the strength of evidence contributed by the sampling plan in relation to the study’s design. Each inclusion or exclusion criterion should have a rationale, presumably related to a potential contaminating effect on the dependent variable. For example, subjects were excluded from this study if they had:



The careful establishment of sample inclusion or exclusion criteria will increase the study’s precision and strength of evidence, thereby contributing to the accuracy and generalizability of the findings (see Chapter 8).







Samples and sampling


Sampling is a process of selecting a portion or subset of the designated population to represent the entire population. A sample is a set of elements that make up the population; an element is the most basic unit about which information is collected. The most common element in nursing research is individuals, but other elements (e.g., places, objects) can form the basis of a sample or population. For example, a researcher was planning a study that compared the effectiveness of different nursing interventions on reducing falls in the elderly in long-term care facilities (LTCs). Four LTCs, each using a different falls prevention treatment protocol, were identified as the sampling units rather than the nurses themselves or the treatment alone.


The purpose of sampling is to increase a study’s efficiency. As a new evaluator of research you must realize that it would not be feasible to examine every element in the population. When sampling is done properly, the researcher can draw inferences and make generalizations about the population without examining each element in the population. Sampling procedures that formulate specific criteria for selection ensure that the characteristics of the phenomena of interest will be, or are likely to be, present in all of the units being studied. The researcher’s efforts to ensure that the sample is representative of the target population strengthens the evidence generated by the sample composition, which puts the researcher in a stronger position to draw conclusions that are generalizable to the population and applicable to practice (see Chapter 8).


After having reviewed a number of research studies, you will recognize that samples and sampling procedures vary in terms of merit. The foremost criterion in appraising a sample is its representativeness. A representative sample is one whose key characteristics closely approximate those of the population. If 70% of the population in a study of child-rearing practices consisted of women and 40% were full-time employees, a representative sample should reflect these characteristics in the same proportions.





Types of samples


Sampling strategies are generally grouped into two categories: nonprobability sampling and probability sampling. In nonprobability sampling, elements are chosen by nonrandom methods. The drawback of this strategy is that there is no way of estimating each element’s probability of being included in a particular sample. Essentially, there is no way of ensuring that every element has a chance for inclusion in the nonprobability sample.


Probability sampling uses some form of random selection when the sample is chosen. This type of sample enables the researcher to estimate the probability that each element of the population will be included in the sample. Probability sampling is the more rigorous type of sampling strategy and is more likely to result in a representative sample. The remainder of this section is devoted to a discussion of different types of nonprobability and probability sampling strategies. A summary of sampling strategies appears in Table 12-1. You may wish to refer to this table as the various nonprobability and probability strategies are discussed in the following sections.






Nonprobability sampling


Because of lack of randomization, the findings of studies using nonprobability sampling are less generalizable than those using a probability sampling strategy, and they tend to produce fewer representative samples. Such samples are easier for the researcher to obtain, however, and many samples—not only in nursing research, but also in other disciplines—are nonprobability samples. When a nonprobability sample is carefully chosen to reflect the target population through the careful use of inclusion and exclusion criteria and adequate sample size, you can have more confidence in the sample’s representativeness and the external validity of the findings. The three major types of nonprobability sampling are convenience, quota, and purposive sampling strategies.



Convenience sampling

Convenience sampling is the use of the most readily accessible persons or objects as subjects. The subjects may include volunteers, the first 100 patients admitted to hospital X with a particular diagnosis, all of the people enrolled in program Y during the month of September, or all of the students enrolled in course Z at a particular university during 2014. The subjects are convenient and accessible to the researcher and are thus called a convenience sample. For example, a researcher studying the relationship among maternal-fetal attachment (MFA), health practices during pregnancy, and neonatal outcomes in a sample of low-income predominantly African-American women and their neonates recruited a convenience sample of 167 women from three urban obstetrical clinics in the Mid Atlantic region who met the eligibility criteria and volunteered to participate in the study (Alhusen et al., 2012; see Appendix B).


The advantage of a convenience sample is that generally it is easier for the researcher to obtain subjects. The researcher may have to be concerned only with obtaining a sufficient number of subjects who meet the same criteria. A convenience sample may be the most appropriate sampling strategy to use even though it is not the strongest approach. The major disadvantage of a convenience sample is that the risk of bias is greater than in any other type of sample (see Table 12-1). The fact that convenience samples use voluntary participation increases the probability of researchers recruiting those people who feel strongly about the issue being studied, which may favor certain outcomes (Sousa et al., 2004). In this case, you can ask yourself the following as you think about the strength and quality of evidence contributed by the sampling component of a study:



Researchers may recruit subjects when they stop people on a street corner to ask their opinion on some issue, place advertisements in the newspaper, or place signs in local churches, community centers, or supermarkets indicating that volunteers are needed for a particular study. To assess the degree to which a convenience sample approximates a random sample, the researcher checks for the representativeness of the convenience sample by comparing the sample to population percentages and, in that way, assesses the extent to which bias is or is not evident (Sousa et al., 2004).


Because acquiring research subjects is a problem that confronts many nurse researchers, innovative recruitment strategies may be used. A unique method of accessing and recruiting subjects is the use of online computer networks (e.g., disease-specific chat rooms, blogs, and bulletin boards). In the evidence hierarchy in Figure 1-1, nonprobability sampling is most commonly associated with quantitative nonexperimental or qualitative studies that contribute Level IV through Level VI evidence.



When you appraise a study you should recognize that the convenience sample strategy, although the most common, is the weakest sampling strategy with regard to strength of evidence and generalizability. When a convenience sample is used, caution should be exercised in interpreting the data. When critiquing a study that has employed this sampling strategy, the reviewer should be justifiably skeptical about the external validity and applicability of the findings (see Chapter 8).



Quota sampling

Quota sampling refers to a form of nonprobability sampling in which knowledge about the population of interest is used to build some representativeness into the sample (see Table 12-1). A quota sample identifies the strata of the population and proportionally represents the strata in the sample. For example, the data in Table 12-2 reveal that 40% of the 5000 nurses in city X are associate degree graduates, 30% are 4-year baccalaureate degree graduates, and 30% are accelerated baccalaureate graduates. Each stratum of the population should be proportionately represented in the sample. In this case, the researcher used a proportional quota sampling strategy and decided to sample 10% of a population of 5000 (i.e., 500 nurses). Based on the proportion of each stratum in the population, 400 associate degree graduates, 300 4-year baccalaureate graduates, and 300 accelerated baccalaureate graduates were the quotas established for the three strata. The researcher recruited subjects who met the study’s eligibility criteria until the quota for each stratum was filled. In other words, once the researcher obtained the necessary 400 associate degree graduates, 300 4-year baccalaureate degree graduates, and 300 accelerated baccalaureate degree graduates, the sample was complete.


Feb 15, 2017 | Posted by in NURSING | Comments Off on Sampling

Full access? Get Clinical Tree

Get Clinical Tree app for offline access