Examining Populations and Samples in Research



Examining Populations and Samples in Research




Students often enter the field of research with preconceived notions about samples and sampling methods. Many of these notions come from exposure to television advertisements, public opinion polls, and newspaper reports of research findings. A television spokesperson boasts that four of five doctors recommend a particular pain medication, a newscaster announces that John Jones will win the senate election by a margin of 10%, and a newspaper reporter writes that research has shown that aggressive treatment of hypertension to maintain a blood pressure of 120/80 mm Hg or lower significantly reduces the risk for coronary artery disease and stroke.


All these examples include a sampling technique or method. Some of the outcomes from these sampling methods are more valid than others, based on the sampling method used and the sample size achieved. When critically appraising a study, you need to examine the sampling process and determine its quality. The sampling process is usually described in the methods section of a published research report. This chapter was developed to assist you in understanding and critically appraising the sampling processes implemented in quantitative, qualitative, and outcomes studies. Initially, the concepts of sampling theory are introduced, including sampling criteria, sampling frame, and representativeness of a sample. The nonprobability and probability sampling methods and sample sizes for quantitative and qualitative studies are detailed. The chapter concludes with a discussion of the natural, partially controlled, and highly controlled settings used in conducting research.



Understanding Sampling Concepts


Sampling involves selecting a group of people, events, objects, or other elements with which to conduct a study. A sampling method or plan defines the selection process, and the sample defines the selected group of people (or elements). A sample selected in a study should represent an identified population of people. The population might be all people who have diabetes, all patients who have had abdominal surgery, or all persons who receive care from a registered nurse. In most cases, however, it would be impossible for researchers to study an entire population. Sampling theory was developed to determine the most effective way to acquire a sample that accurately reflects the population under study. Key concepts of sampling theory include populations, target population, sampling or eligibility criteria, accessible population, elements, representativeness, sampling frames, and sampling methods or plans. The following sections describe these concepts and include relevant examples from published studies.



Populations and Elements


The population is a particular group of individuals or elements, such as people with type 2 diabetes, who are the focus of the research. The target population is the entire set of individuals or elements who meet the sampling criteria (defined in the next section), such as female, 18 years of age or older, new diagnosis of type 2 diabetes confirmed by the medical record, and not on insulin. Figure 9-1 demonstrates the link of the population, target population, and accessible population in a study. An accessible population is the portion of the target population to which the researcher has reasonable access. The accessible population might include elements within a country, state, city, hospital, nursing unit, or primary care clinic, such as the individuals with diabetes who were provided care in a primary care clinic in Arlington, Texas. Researchers obtain the sample from the accessible population by using a particular sampling method or plan, such as simple random sampling. The individual units of the population and sample are called elements. An element can be a person, event, object, or any other single unit of study. When elements are persons, they are referred to as participants or subjects (see Figure 9-1). Quantitative and outcomes researchers refer to the people they study as subjects or participants. Qualitative researchers refer to the individuals they study as participants.



Generalization extends the findings from the sample under study to the larger population. In quantitative and outcomes studies, researchers obtain a sample from the accessible population with the goal of generalizing the findings from the sample to the accessible population and then, more abstractly, to the target population (see Figure 9-1). The quality of the study and consistency of the study’s findings with the findings from previous research in this area influence the extent of the generalization. If a study is of high quality, with findings consistent with previous research, then researchers can be more confident in generalizing their findings to the target population. For example, the findings from the study of female patients with a new diagnosis of type 2 diabetes in a primary care clinic in Arlington, Texas, may be generalized to the target population of women with type 2 diabetes managed in primary care clinics. With this information, you can decide whether it is appropriate to use this evidence in caring for the same type of patients in your practice, with the goal of moving toward evidence-based practice (EBP; Brown, 2014; Melnyk & Fineout-Overholt, 2011).



Sampling or Eligibility Criteria


Sampling or eligibility criteria include the list of characteristics essential for eligibility or membership in the target population. For example, researchers may choose to study the effect of preoperative teaching about early ambulation on the outcome of length of hospital stay for adults having knee joint replacement surgery. In this study, the sampling criteria may include (1) age of at least 18 years of age or older (adults), (2) able to speak and read English, (3) surgical replacement of one knee joint, (4) no history of previous joint replacement surgery, (5) no diagnosis of dementia, and (6) no debilitating chronic muscle diseases. The sample is selected from the accessible population that meets these sampling criteria. Sampling criteria for a study may consist of inclusion or exclusion sampling criteria, or both. Inclusion sampling criteria are the characteristics that the subject or element must possess to be part of the target population. In the example, the inclusion criteria are age 18 years of age or older, able to speak and read English, and surgical replacement of one knee joint. Exclusion sampling criteria are those characteristics that can cause a person or element to be excluded from the target population. For example, any subjects with a history of previous joint replacement surgery, diagnosis of dementia, and diagnosis of a debilitating chronic muscle disease were excluded from the preoperative teaching study. Researchers should state a sample criterion only once and should not include it as both an inclusion and exclusion criterion. Thus, researchers should not have an inclusion criterion of no diagnosis of dementia and an exclusion criterion of diagnosis of dementia.


When the quantitative or outcomes study is completed, the findings are often generalized from the sample to the target population that meets the sampling criteria (Fawcett & Garity, 2009). Researchers may narrowly define the sampling criteria to make the sample as homogeneous (or similar) as possible to control for extraneous variables. Conversely, the researcher may broadly define the criteria to ensure that the study sample is heterogeneous, with a broad range of values or scores on the variables being studied. If the sampling criteria are too narrow and restrictive, researchers may have difficulty obtaining an adequately sized sample from the accessible population, which can limit the generalization of findings.


In discussing the generalization of quantitative study findings in a published research report, investigators sometimes attempt to generalize beyond the sampling criteria. Using the example of the early ambulation preoperative teaching study, the sample may need to be limited to subjects who speak and read English because the preoperative teaching is in English and one of the measurement instruments requires that subjects be able to read English. However, the researchers may believe that the findings can be generalized to non–English-speaking persons. When reading studies, you need to consider carefully the implications of using these findings with a non–English-speaking population. Perhaps non–English-speaking persons, because they come from another culture, do not respond to the teaching in the same way as that observed in the study population. When critically appraising a study, examine the sample inclusion and exclusion criteria, and determine whether the generalization of the study findings is appropriate based on the study sampling criteria. (Chapter 11 provides more detail on generalizing findings from studies.)



Representativeness of a Sample in Quantitative and Outcomes Research


Representativeness means that the sample, accessible population, and target population are alike in as many ways as possible (see Figure 9-1). In quantitative and outcomes research, you need to evaluate representativeness in terms of the setting, characteristics of the subjects, and distribution of values on variables measured in the study. Persons seeking care in a particular setting may be different from those who seek care for the same problem in other settings or those who choose to use self-care to manage their problems. Studies conducted in private hospitals usually exclude low-income patients. Other settings may exclude older adults or those with less education. People who do not have access to care are usually excluded from studies. Subjects in research centers and the care that they receive are different from patients and the care that they receive in community hospitals, public hospitals, veterans’ hospitals, or rural hospitals. People living in rural settings may respond differently to a health situation from those who live in urban settings. Thus the setting identified in published studies does influence the representativeness of the sample. Researchers who gather data from subjects across a variety of settings have a more representative sample of the target population than those limiting the study to a single setting.


A sample must be representative in terms of characteristics such as age, gender, ethnicity, income, and education, which often influence study variables. These are examples of demographic or attribute variables that might be selected by researchers for examination in their study. Researchers analyze data collected on the demographic variables to produce the sample characteristics—characteristics used to provide a picture of the sample. These sample characteristics must be reasonably representative of the characteristics of the population. If the study includes groups, the subjects in the groups must have comparable demographic characteristics (see Chapter 5 for more details on demographic variables and sample characteristics).


Studies that obtain data from large databases have more representative samples. For example, Monroe, Kenaga, Dietrich, Carter, and Cowan (2013) examined the prevalence of employed nurses enrolled in substance use monitoring programs by examining data from the National Council of State Boards of Nursing (NCSBN) 2010 Survey of Regulatory Boards Disciplinary Actions on Nurses. This NCSBN survey included the United States and its territories and found that 17,085 (0.51%) of the employed nurses were enrolled in substance use monitoring programs. This study examined data from multiple sites (United States and its territories) and included a large national population of nurses (all employed nurses), resulting in a representative sample.



Random and Systematic Variation of Subjects’ Values


Measurement values also need to be representative. Measurement values in a study often vary randomly among subjects. Random variation is the expected difference in values that occurs when different subjects from the same sample are examined. The difference is random because some values will be higher and others lower than the average (mean) population value. As sample size increases, random variation decreases, improving representativeness.


Systematic variation, or systematic bias—a serious concern in sampling—is a consequence of selecting subjects whose measurement values differ in some specific way from those of the population. This difference usually is expressed as a difference in the average (or mean) values between the sample and population. Because the subjects have something in common, their values tend to be similar to those of others in the sample but different in some way from those of the population as a whole. These values do not vary randomly around the population mean. Most of the variation from the mean is in the same direction; it is systematic. Thus the sample mean may be higher than or lower than the mean of the target population. Increasing the sample size has no effect on systematic variation. For example, if all the subjects in a study examining some type of knowledge level have an intelligence quotient (IQ) above 120, then all their test scores in the study are likely to be higher than those of the population mean, which includes people with a wide variation in IQ scores (but with a mean IQ of 100). The IQs of the subjects will introduce a systematic bias. When systematic bias occurs in quasi-experimental or experimental studies, it can lead the researcher to conclude that the treatment has made a difference, when in actuality the values would have been different, even without the treatment.



Acceptance and Refusal Rates in Studies


The probability of systematic variation increases when the sampling process is not random. Even in a random sample, however, systematic variation can occur when a large number of the potential subjects declines participation. As the number of subjects declining participation increases, the possibility of a systematic bias in the study becomes greater. In published studies, researchers may identify a refusal rate, which is the percentage of subjects who declined to participate in the study, and the subjects’ reasons for not participating (Grove, Burns, & Gray, 2013). The formula for calculating the refusal rate in a study is as follows:


Refusalrate=Numberrefusingparticipation÷numbermeetingsamplingcriteriaapproached×100%



image


For example, if 80 potential subjects meeting sampling criteria are approached to participate in the hypothetical study about the effects of early ambulation preoperative teaching on length of hospital stay, and 4 patients refuse, then the refusal rate would be:


Refusalrate=4÷80×100%=0.5×100%=5%



image


Other studies record an acceptance rate, which is the percentage of subjects meeting sampling criteria consenting to participate in a study. However, researchers will report the refusal or acceptance rate, but not both. The formula for calculating the acceptance rate in a study is as follows:


Acceptancerate=Numberacceptingparticipation÷numbermeetingsamplingcriteriaapproached×100%



image


In the hypothetical preoperative teaching study, 4 of 80 potential subjects refused to participate—so 80 − 4 = 76 accepted. Plugging the following numbers into the stated formula gives:


Acceptancerate=76÷80×100%=0.95×100%=95%



image


You can also calculate the acceptance and refusal rates as follows:


Acceptancerate=100%refusalrate



image


Or:


Refusalrate=100%acceptancerate



image


In this example, the acceptance rate was 100%  5% (refusal rate) = 95%, which is high or strong. In studies with a high acceptance rate or a low refusal rate reported, the chance for systematic variation is less, and the sample is more likely to be representative of the target population. Researchers usually report the refusal rate, and it is best to provide rationales for the individuals refusing to participate.



Sample Attrition and Retention Rates in Studies


Systematic variation also may occur in studies with high sample attrition. Sample attrition is the withdrawal or loss of subjects from a study that can be expressed as a number of subjects withdrawing or a percentage. The percentage is the sample attrition rate and it is best if researchers include both the number of subjects withdrawing and the attrition rate. The formula for calculating the sample attrition rate in a study is as follows:

Sampleattritionrate=Numberofsubjectswithdrawingfromastudy÷samplesizeofstudy×100%image

For example, in the hypothetical study of preoperative teaching, 31 subjects—12 from the treatment group and 19 from the comparison group—withdraw, for various reasons. Loss of 31 subjects means a 41% attrition rate:


Sampleattritionrate=31÷76×100%=0.418×100%=40.8%=41%



image


In this example, the overall sample attrition rate was considerable (41%), and the rates differed for the two groups to which the subjects were assigned. You can also calculate the attrition rates for the groups. If the two groups were equal at the start of the study and each included 38 subjects, then the attrition rate for the treatment group was (12 ÷ 38) × 100% = 0.316 × 100% = 31.6% = 32%. The attrition for the comparison group was (19 ÷ 38) x 100% = 0.5 × 100% = 50%. Systematic variation is greatest when a large number of subjects withdraw from the study before data collection is completed or when a large number of subjects withdraw from one group but not the other(s) in the study. In studies involving a treatment, subjects in the comparison group who do not receive the treatment may be more likely to withdraw from the study. However, sometimes the attrition is higher for the treatment group if the intervention is complex and/or time-consuming (Kerlinger & Lee, 2000). In the early ambulation preoperative teaching example, there is a strong potential for systematic variation because the sample attrition rate was large (41%) and the attrition rate in the comparison group (50%) was larger than the attrition rate in the treatment group (32%). The increased potential for systematic variation results in a sample that is less representative of the target population.


The opposite of sample attrition is the sample retention, which is the number of subjects who remain in and complete a study. You can calculate the sample retention rate in two ways:


Sampleretentionrate=Numberofsubjectscompletingthestudy÷samplesize×100%



image


Or:


Sampleretentionrate=100%sampleattritionrate



image


In the example, early ambulation preoperative teaching study, 45 subjects were retained in the study that had an original sample of 76 subjects:


Sampleretentionrate=45÷76×100%=0.59×100%=59.2%=59%



image


Or:


Sampleretentionrate=100%41%=59%



image


The higher the retention rate, the more representative the sample is of the target population and the more likely the study results are an accurate reflection of reality. Often, researchers will identify the attrition rate or retention rate, but not both. It is best to provide a rate in addition to the number of subjects withdrawing from a study, as well as the subjects’ reasons for withdrawing.



Sampling Frames


From a sampling theory perspective, each person or element in the population should have an opportunity to be selected for the sample. One method of providing this opportunity is referred to as random sampling. For everyone in the accessible population to have an opportunity for selection in the sample, each person in the population must be identified. To accomplish this, the researcher must acquire a list of every member of the population, using the sampling criteria to define eligibility. This list is referred to as the sampling frame. In some studies, the complete sampling frame cannot be identified because it is not possible to list all members of the population. The Health Insurance Portability and Accountability Act (HIPAA) has also increased the difficulty in obtaining a complete sampling frame for many studies because of its requirements to protect individuals’ health information (see Chapter 4 for more information on HIPAA). Once a sampling frame is identified, researchers select subjects for their studies using a sampling plan or method.



Sampling Methods or Plans


Sampling methods or plans outline strategies used to obtain samples for studies. Like a design, a sampling plan is not specific to a study. The sampling plan may include probability (random) or nonprobability (nonrandom) sampling methods. Probability sampling methods are designed to




image Research Example


Sampling Criteria, Acceptance or Refusal Rate, and Sample Attrition or Retention Rate



Research Excerpt


Giakoumidakis and colleagues (2013) conducted a randomized controlled trial (RCT) to investigate the effects of intensive blood glucose control (120 to 150 mg/dL) on cardiac surgery patient outcomes, such as mortality (in-hospital and 30-day postdischarge), length of intensive care unit (ICU) stay, length of postoperative hospital stay, duration of tracheal intubation, presence of severe hypoglycemic events, and incidence of postoperative infections. The sampling criteria, acceptance rate, and sample retention rate from this study are presented here as an example. Giakoumidakis and associates (2013) provided a description of their study sampling criteria and documented the participants enrolled in their study using a flow diagram (see Figure 9-2 of this example). This flow diagram is based on the CONsolidated Standards of Reporting Trials (CONSORT) Statement that is the international standard for reporting the sampling process in RCTs (CONSORT Group, 2010; see Chapter 13 for more information on the CONSORT statement).This study is critically appraised using the questions designated earlier in the Critical Appraisal Guidelines box.




“The study was a randomized quasi-experimental trial. We treated blood glucose levels during the first 24 hours postoperatively [independent variable] in patients of the therapy group and compared them with the control group. The inclusion criteria were: (1) open heart surgery, (2) surgery requiring CPB [cardio-pulmonary bypass], (3) patient age ≥ 18 years old, and (4) the patient’s informed consent for participation in our study. The exclusion criteria included: (1) renal dysfunction or failure (preoperative creatinine > 1.5 mg/dL), (2) neurological or mental disorder, (3) chronic obstructive pulmonary disease, (4) preoperative use of any type of antibiotics, (5) emergency and urgent surgeries, (6) history of previous cardiac surgery, (7) ICU length of stay < 24 hours, (8) mediastinal re-exploration for bleeding, (9) hemodynamic support with intra-aortic balloon pump (IABP) intraoperatively and/or during the first 24 hours postoperatively, and (10) use of cardioversion for severe ventricular arrhythmias (ventricular tachycardia and/or fibrillation) within the first 24 hours of ICU hospitalization. These criteria were established in an effort to ensure a more homogenous sample for our study.…


Over a period of 5 months (from September 2011 to January 2012), 298 patients were admitted to the 8-bed cardiac surgery ICU… and were eligible for enrollment in the study. Two hundred and twelve out of 298 (71.1%) patients met the inclusion criteria and simultaneously did not meet the exclusion criteria and consequently constituted our study sample (see Figure 9-2).


One of the researchers, the same each time, randomly assigned patients, immediately postoperatively, to the (odd numbers into the control group and the evens into the therapy group): (1) control group (n = 107) with a targeted blood glucose levels of 161-200 mg/dL, or (2) therapy group (n = 105) with blood glucose target of 120-160 mg/dL, or during the first 24 hours postoperatively. (see Figure 9-2).”


(Giakoumidakis et al., 2013, p. 147)



Critical Appraisal


Giakoumidakis and co-workers (2013) identified specific inclusion and exclusion sampling criteria to designate the subjects in the target population selectively. As the researchers indicated, the sampling criteria were narrowly defined by the researchers to promote the selection of a homogeneous sample of cardiac surgery patients. These sampling criteria were appropriate for this study to reduce the effects of possible extraneous variables on the implementation of the treatment (blood glucose control) and the measurement of the dependent variables or outcomes (mortality, length of ICU and hospital stays, duration of tracheal intubation, presence of severe hypoglycemic events, and incidence of postoperative infection). The increased controls imposed by the sampling criteria strengthened the likelihood that the study outcomes were caused by the treatment and not by extraneous variables.


These researchers assessed 298 patients for eligibility in the study, but 83 of them did not meet the sampling criteria. Thus 215 patients met the sampling criteria. However, three of these patients declined or refused to participate in the study, resulting in a 1.4% (3 ÷ 215 × 100% = 0.014 × 100% = 1.4%) refusal rate. Figure 9-2 indicates that the sample of 212 patients was equally randomized into the control group (n = 107) and therapy group (n = 105). There was no attrition of subjects from this study, as indicated in Figure 9-2, in which the starting group sizes were the same at the end and all subjects (N = 212) were included in the data analyses. This study had very rigorous sampling criteria, low refusal rate of 1.4% (acceptance rate of 98.6%), and 0% attrition rate, which increased the representativeness of the sample of the accessible and target populations. The study would have been strengthened by the researchers including not only the numbers but also the sample refusal rate and the reasons for the three patients refusing to participate in the study.



Implications for Practice


Giakoumidakis and colleagues (2013) found that only the in-hospital mortality rate was significantly affected by the intensive blood glucose control. The postoperative glycemic control did not affect the other patient outcomes that were studied. However, the researchers recognized that the sample size was small for a RCT, which limited the numbers of patients in the control and therapy groups. In addition, the study was conducted using only the patients from one hospital. Therefore the researchers recommended that future studies include larger samples obtained from a variety of hospitals. This initial research indicates that intensive blood glucose control is important to cardiac patients’ mortality and might have other beneficial effects on patient outcomes in future research. The Quality and Safety Education for Nurses (QSEN) importance is that these research findings provide knowledge to address patient-centered care and safety competencies to promote quality care to patients and families (QSEN, 2013; Sherwood & Barnsteiner, 2012).


increase representativeness and decrease systematic variation or bias in quantitative and outcomes studies. When critically appraising a study, identify the study sampling plan as probability or nonprobability, and determine the specific method or methods used to select the sample. The different types of probability and nonprobability sampling methods are introduced next.



Probability Sampling Methods


In probability sampling, each person or element in a population has an opportunity to be selected for a sample, which is achieved through random sampling. Probability or random sampling methods increase the sample’s representativeness of the target population. All the subsets of the population, which may differ from each other but contribute to the parameters (e.g., the means and standard deviations) of the population, have a chance to be represented in the sample. The opportunity for systematic bias is less when subjects are selected randomly, although it is possible for a systematic bias to occur by chance.


Without random sampling strategies, researchers, who have a vested interest in the study, might tend (consciously or unconsciously) to select subjects whose conditions or behaviors are consistent with the study hypotheses. For example, researchers may exclude potential subjects because they are too sick, not sick enough, coping too well, not coping adequately, uncooperative, or noncompliant. By using random sampling, however, researchers leave the selection to chance, thereby increasing the validity of their study findings.


There are four sampling designs that achieve probability sampling included in this text—simple random sampling, stratified random sampling, cluster sampling, and systematic sampling. Table 9-1 identifies the common probability and nonprobability sampling methods used in nursing studies, their applications, and their representativeness for the study. Probability and nonprobability sampling methods are used in quantitative and outcomes studies, and nonprobability sampling methods are used in qualitative studies (Fawcett & Garity, 2009; Munhall, 2012).



Table 9-1


Probability and Nonprobability Sampling Methods

















































Sampling Method Common Application(s) Representativeness
Probability
Simple random sampling Quantitative and outcomes research Strong representativeness of the target population that increases with sample size
Stratified random sampling Quantitative and outcomes research Strong representativeness of the target population that increases with control of stratified variable(s)
Cluster sampling Quantitative and outcomes research Less representative of the target population than simple random sampling and stratified random sampling
Systematic sampling Quantitative and outcomes research Less representative of the target population than simple random sampling and stratified random sampling methods
Nonprobability
Convenience sampling Quantitative, qualitative, and outcomes research Questionable representativeness of the target population that improves with increasing sample size; may be representative of the phenomenon, process, or cultural elements in qualitative research
Quota sampling Quantitative and outcomes research and, rarely, qualitative research Use of stratification for selected variables in quantitative research makes the sample more representative than convenience sampling.
In qualitative research, stratification might be used to provide greater understanding and increase the representativeness of the phenomenon, processes, or cultural elements.
Purposeful or purposive sampling Qualitative and sometimes quantitative research Focus is on insight, description, and understanding of a phenomenon or process with specially selected study participants.
Network or snowball sampling Qualitative and sometimes quantitative research Focus is on insight, description, and understanding of a phenomenon or process in a difficult to access population.
Theoretical sampling Qualitative research Focus is on developing a theory in a selected area.

Stay updated, free articles. Join our Telegram channel

Feb 9, 2017 | Posted by in NURSING | Comments Off on Examining Populations and Samples in Research

Full access? Get Clinical Tree

Get Clinical Tree app for offline access