This chapter describes the designs most commonly used in nursing research, using the study categories described in Chapter 3: descriptive, correlational, quasi-experimental, and experimental. Descriptive and correlational designs examine variables in natural environments, such as home, and do not include researcher-designed treatments or interventions. Quasi-experimental and experimental designs examine the effects of an intervention by comparing differences between groups that have received the intervention and those that have not received the intervention. As you review each design, note the threats to validity that are controlled by the design, keeping in mind that uncontrolled threats in the design you choose may weaken the validity of your study. Box 11-1 lists the designs discussed in this chapter. Each of the designs is briefly described, and a model is provided so you can see the different elements of the designs. After the descriptions of the designs, we provide a series of decision trees or algorithms that will help you to select the appropriate design for your study or to identify the design used in a published study. Originally, only experimental designs were considered of value. In addition, many believed that the only setting in which an experiment can be conducted is a laboratory, where stricter controls can be maintained than in a field or natural setting. This approach is appropriate for the natural sciences but not for the social sciences. From the social sciences have emerged additional quantitative designs (descriptive, correlational, and quasi-experimental), methodological designs, and qualitative designs (Cook & Campbell, 1979; Creswell, 2009; Fawcett & Garity, 2009). The epidemiology, public health, and community health fields have presented time-series designs, health promotion designs, and prevention designs. Innovative design strategies are beginning to appear within nursing research. One example is the intervention research design described in Chapter 14. Developing designs to study the outcomes of nursing actions is also important. The emerging field of outcomes research in nursing is described in Chapter 13. Nurse researchers must see themselves as credible scientists before they will dare to develop new design strategies to explore little-understood aspects of nursing. To develop a new design, the researcher must carefully consider possible threats to validity and ways to diminish them. Nurses must also be willing to risk the temporary failures that are always inherent in the development of something new. Descriptive designs vary in levels of complexity. Some contain only two variables, whereas others may have multiple variables. The relationships among variables present an overall picture of the phenomenon being examined, but examination of types and degrees of relationships is not the primary purpose of a descriptive study. Protection against bias (or threat to the validity) in a descriptive design is achieved through (1) links between conceptual and operational definitions of variables (Fawcett & Garity, 2009); (2) sample selection and size (Aberson, 2010; Thompson, 2002); (3) the use of valid and reliable instruments (Waltz, Strickland, & Lenz, 2010) or accurate and precise biophysical measures (Ryan-Wenger, 2010); and (4) data collection procedures that achieve some environmental control (Bialocerkowski, Klupp, & Bragge, 2010; Creswell, 2009; DeVon et al., 2007; Kerlinger & Lee, 2000). It is not uncommon for researchers using a descriptive design to combine quantitative descriptive methods and qualitative methods (mixed methods) (Creswell, 2009). Chapter 10 includes a discussion of different types of mixed-methods approaches. To use this strategy, consult with a researcher experienced in using qualitative methods or include this person as a research partner to appropriately collect, analyze, and interpret qualitative data. The comparative descriptive design (Figure 11-2) examines and describes differences in variables in two or more groups that occur naturally in the setting. Descriptive statistics and inferential statistical analyses may be used to examine differences between or among groups. Commonly, the results obtained from these analyses are not generalized to a population because the description is for a very specific sample and would not necessarily apply to a larger population. An example of this design is the study by Cramer, Chen, Roberts, and Clute (2007) of the social and economic impact of community-based prenatal care. The abstract for this study, which is reprinted in full, describes the focus, design, and major findings: Because of the large volumes of data acquired in a longitudinal study, you must give careful attention to strategies for managing the data. The repetition of measures requires that data analysis be carefully thought through. Analyses commonly used are repeated measures analyses of variance, multivariate analyses of variance (MANOVA), regression analysis, cluster analysis, and time-series analysis (see Chapters 24 and 25) (Corty, 2007; Munro, 2005). Trend designs examine changes in the general population in relation to a particular phenomenon (see Figure 11-5). The researchers select different samples of subjects from the same population at preset intervals of time, and at each selected time, they collect data from that particular sample. Researchers need to be able to justify generalizing from the samples to the population under study. Analysis involves strategies to predict future trends by examining past trends. Harris, Gordon-Larsen, Chantala, and Udry (2006, p. 74) used a trend design to describe “longitudinal trends in race/ethnic disparities in 20 leading health indicators from Healthy People 2010 [U.S. Department of Health and Human Services, 2000] across multiple domains from adolescence to young adulthood.” These researchers examined the study trends in an ethnically diverse, national database, and their study design is described in the following excerpt. A merger of the cross-sectional or longitudinal and trend designs, the event-partitioning design, is used in some cases to enlarge sample size and to avoid the effects of history on the validity of findings. Cook and Campbell (1979) referred to these as cohort designs with treatment partitioning. Figure 11-6 shows a model of the cross-sectional study design with treatment partitioning, and Figure 11-7 provides the model of a longitudinal design with treatment partitioning. The term treatment is used loosely here to mean a key event that is thought to lead to change. In a descriptive study, the researcher would not cause or manipulate the key event but rather would clearly define it so that when it occurred naturally, it would be recognized. Case studies were commonly used in nursing research in the 1970s. Their use then declined, but they are beginning to appear in the literature more frequently today. Well-designed case studies are good sources of descriptive information and can be used as evidence for or against theories. Case studies can use a mixed-methods approach, incorporating both quantitative and qualitative methods (Creswell, 2009; Fawcett & Garity, 2009). Sterling and McNally (1992) recommended single-subject case studies for examining process-based nursing practice. This strategy allows the researcher to investigate daily observations and interventions that are a common aspect of nursing practice. How you design a case study depends on the circumstances of the case but usually includes an element of time. History and previous behavior patterns are usually explored in detail. As the case study proceeds, you may become aware of components important to the phenomenon being examined that were not originally built into the study. A case study is likely to have both quantitative and qualitative elements; and if the study incorporates both of these components, the study design must clearly present this fact (Creswell, 2009). Methods used to analyze and interpret qualitative data need to be carefully planned. Consultation with a qualitative researcher can strengthen the study. Large volumes of data are generally obtained during a case study. Organizing the findings of a case study into a coherent whole is a difficult but critical component of the study (Fawcett & Garity, 2009). Generalizing study findings in the statistical sense is not appropriate; however, generalizing the findings to theory is appropriate and important (Crombie & Davies, 1996; Gray, 1998; Sandelowski, 1996). Neophyte researchers tend to make two serious errors with correlational studies. First, they often attempt to establish causality by correlation, reasoning that if two variables are related, one must cause the other. Second, they confuse studies in which differences are examined with studies in which relationships are examined. Although the existence of a difference assumes the existence of a relationship, the design and statistical analysis of studies examining differences are not the same as those of studies examining relationships. If your study examines two or more groups in terms of one or more variables, then you are exploring differences between or among groups as reflected in scores on the identified variables. If your study examines a single group in terms of two or more variables, then you are exploring relationships between or among variables. In a correlational study, the relationship examined is between or among two or more research variables within an identified situation. Thus, the sample is not separated into groups. Analyses examine variable values in the entire sample. In a correlational design, data from the entire sample are analyzed as a single group (Grove, 2007; Kerlinger & Lee, 2000). This study had a descriptive correlational design, as evidenced by the single study group, the absence of treatment, and the use of descriptive and correlational statistical techniques to analyze study data. The study variables informational support, anxiety, and satisfaction with care were described with means, standard deviations, and ranges (see Chapter 22). The relationships among these three variables were determined using Pearson’s product moment correlational coefficient (see Chapter 23). The aim of a predictive design is to predict the level of the dependent variable from the independent variables. Figure 11-9 is a model of a predictive design with two independent variables used to predict the dependent variable. Independent variables most effective in prediction are highly correlated with the dependent variable but not highly correlated with other independent variables used in the study. Predictive designs require you to develop a theory-based mathematical hypothesis proposing the independent variables that are expected to predict the dependent variable effectively. You can then test the hypothesis using regression analysis (see Chapter 24) (Corty, 2007; Munro, 2005). Predictive studies are also used to establish the predictive validity of measurement scales. Variables are classified into three categories: exogenous variables, endogenous variables, and residual variables. Exogenous variables are within the theoretical model but are caused by factors outside this model. Endogenous variables are those whose variation is explained within the theoretical model. Exogenous variables influence the variation of endogenous variables. Residual variables indicate the effect of unmeasured variables not included in the model. These variables explain some of the variance found in the data but not the variance within the model (Mason-Hawkes & Holm, 1989; Norris, 2005a). To measure exogenous and endogenous variables, you would collect data from the subjects and analyze the accuracy of the proposed paths. Historically, these analysis procedures were performed with a series of regression analyses. Researchers now conduct statistical procedures that have been developed specifically for path analysis using the computer programs LISREL and EQS (Norris, 2005a). Structural equation modeling is a commonly used statistical procedure (Norris, 2005b). Path coefficients are calculated that indicate the effect that one variable has on another. The amount of variance explained by the model, as well as the fit between the path coefficients and the theoretical model, indicates the accuracy of the theory. Variance that is not accounted for in the statistical analysis is attributed to residual variables (variables a and b) not included in the analyses (Mason-Hawkes & Holm, 1989; Norris, 2005a). In quasi-experimental and experimental studies, an intervention (or protocol) is developed that is expected to result in differences in posttest measures of the treatment and control or comparison groups. This intervention may be physiological, psychosocial, educational, or a combination of these and should be designed to maximize the differences between the groups. Thus, it should be the best intervention possible in the circumstances of the study and should be expected to improve the outcomes of the experimental group (Egan, Snyder, & Burns, 1992; Forbes, 2009; Santacroce, Maccarelli, & Grey, 2004). Over the last 5 years, the nursing literature has included a growing number of publications focused on the methodology for designing interventions for nursing studies (Morrison et al., 2009; Wyatt, Sikorskii, Rahbar, Victorson, & Adams, 2010; Yamada, Stevens, Sidani, Watt-Watson, & Silva, 2010). In addition, descriptions of nursing interventions in published studies have more detail and specificity but still not at the level given to describing measurement instruments (Fawcett & Garity, 2009; Waltz et al., 2010). Thus, nurse researchers provide detailed information about measurement but often do not provide sufficient detail to allow a nurse to implement a nursing intervention as it was used in a published nursing study. To some extent, this situation may reflect the state of knowledge in the nursing field regarding the provision of nursing interventions in clinical practice. Many clinical nursing interventions are not well defined; thus, each nurse may use her or his own terminology to describe a particular intervention. In addition, an intervention tends to be applied differently in each case by a single nurse and even less consistently by different nurses. However, the quality of nursing interventions has been greatly enhanced with the development of the Nursing Interventions Classification by a team of nurses at the University of Iowa. The Nursing Interventions Classification (NIC) is a standardized language used to describe interventions or treatments performed by nurses in research and practice. Each intervention consists of a label, a definition, and a set of activities performed by nurses carrying out the intervention. The NIC was initiated by the University of Iowa in Iowa City, IA, in 1987 (NIC, 2011). The intervention labels developed over the last 20 years were derived from nursing education and practice. The research to develop the NIC was initiated in 1987 and progressed through four phases that overlapped in time: “Phase I: Construction of the Classification (1987-1992); Phase II: Construction of the Taxonomy (1990-1995); Phase III: Clinical Testing and Refinement (1993-1997); and Phase IV: Use and Maintenance (1996-ongoing)” (Bulechek, Butcher, & Dochterman, 2008, p. 5). The research methods used to develop the classification included content analysis, surveys, focus groups, similarity analysis, and hierarchical clustering. The NIC Taxonomy contained seven domains: Domain 1: Physiological: Basic; Domain 2: Physiological: Complex; Domain 3: Behavioral; Domain 4: Safety; Domain 5: Family; Domain 6: Health System; and Domain 7: Community. There are a total of 30 classes under the seven domains (Bowles & Naylor, 1996; Bulechek et al., 2008). Tripp-Reimer, Woodworth, McCloskey, and Bulechek (1996), in their analysis of the structure of the NIC interventions, identified three dimensions: intensity of care, focus of care, and complexity of care. A high intensity of care is associated with the physiological illness level of the patient and the emergency nature of the illness. The dimension of intensity of care includes indicators of (1) intensity (or acuity) and (2) whether the care is typical or novel. The dimension of focus of care addresses (1) the target of the intervention, ranging from the individual to the system; (2) whether the care action is direct or on behalf of the patient; and (3) the continuum of practice from independent to collaborative actions. The dimension of complexity of care encompasses a range of knowledge, skill, and urgency of the interventions (Bulechek et al., 2008). The interventions in the NIC have been subjected to multiple studies examining the effects on different populations and the effects of varying degrees of intensity. The 5th edition of the Nursing Interventions Classification, developed by faculty at the University of Iowa, included 542 research-based interventions (Bulechek et al., 2008). NIC development continues through the Center for Nursing Classification & Clinical Effectiveness located at the University of Iowa, and you can email them with questions (classification-center@uIowa.edu/; see Chapter 14 for more details on NIC). Currently, studies are being conducted to determine the outcomes of each intervention and to establish links between the intervention and outcomes at varying points in time after the intervention has been implemented. Outcomes that occur immediately following the intervention are easiest to determine. However, the most important outcomes may be those that occur after a client has been discharged or several weeks or months after the intervention. Table 11-1 provides some of the most current examples of the research related to the NIC and the Nursing Outcomes Classification (NOC) being conducted nationally and internationally. This information is critical for ensuring the quality of care provided by nurses and justifying nursing actions in a cost-conscious market (Doran, 2011). For a more extensive discussion of the importance of linking interventions with outcomes measures, see Chapter 13. TABLE 11-1 Work in Nursing Related to the NIC and the Nursing Outcomes Classification (NOC)
Selecting a Quantitative Research Design
evolve.elsevier.com/Grove/practice/
Descriptive Study Designs
Typical Descriptive Study Designs
Comparative Descriptive Designs
Time-Dimensional Designs
Longitudinal Designs
Trend Designs
Event-Partitioning Designs
Case Study Designs
Correlational Study Designs
Descriptive Correlational Designs
Predictive Designs
Model-Testing Designs
Defining Therapeutic Nursing Interventions
The Nursing Interventions Classification
Year
Source
2011
Lee, E., Park, H., Nam, M., & Whyte, J. (2011). Identification and comparison of interventions performed by Korean school nurses and U.S. school nurses using the Nursing Interventions Classification (NIC). Journal of School Nursing, 27(2), 93–101.
2011
Scherb, C. A., Head, B. J., Maas, M. L., Swanson, E. A., Moorhead, S., Reed, D., & Kozel, M. (2011). Most frequent nursing diagnoses, nursing interventions, and nursing-sensitive patient outcomes of hospitalized older adults with heart failure: Part 1. International Journal of Nursing Terminologies & Classifications, 22(1), 13–22.
2010
de Cordova, P., Lucero, R. J., Hyun, S., Quinlan, P., Price, K., & Stone, P. W. (2010). Using the Nursing Interventions Classification as a potential measure of nurse workload. Journal of Nursing Care Quality, 25(1), 39–45.
2010
Lunney, M., McGuire, M., Endozo, N., & McIntosh-Waddy, D. (2010). Consensus-validation study identifies relevant nursing diagnoses, nursing interventions, and health outcomes for people with traumatic brain injuries. Rehabilitation Nursing, 35(4), 161–166.
2010
Smith, K. J., & Craft-Rosenberg, M. (2010). Using NANDA, NIC, and NOC in an undergraduate nursing practicum. Nurse Educator, 35(4), 162–166.
2010
Solari-Twadell, P., & Hackbarth, D. P. (2010). Evidence for a new paradigm of the ministry of parish nursing practice using the nursing intervention classification system. Nursing Outlook, 58(2), 69–75.
2009
Scherb, C. A., & Weydt, A. P. (2009). Work complexity assessment, nursing interventions classification, and nursing outcomes classification: Making connections. Creative Nursing, 15(1), 16–22.
2009
Schneider, J. S., & Slowik, L. H. (2009). The use of the Nursing Interventions Classification (NIC) with cardiac patients receiving home health care. International Journal of Nursing Terminologies & Classifications, 20(3), 132–140.
2009
Wong, E. (2009). Novel nursing terminologies for the rapid response system. International Journal of Nursing Terminologies & Classifications, 20(2), 53–63.
2009
Wong, E., Scott, L. M., Briseno, J. R., Crawford, C. L., & Hsu, J. Y. (2009). Determining critical incident nursing interventions for the critical care setting: A pilot study. International Journal of Nursing Terminologies & Classifications, 20(3), 110–121.
2008
Schneider, J. S., Barkauskas, V., & Keenan, G. (2008). Evaluating home health care nursing outcomes with OASIS and NOC. Journal of Nursing Scholarship, 40(1), 76–82.
2008
Sheerin, F. K. (2008). Diagnoses and interventions pertinent to intellectual disability nursing. International Journal of Nursing Terminologies & Classifications, 19(4), 140–149. You may also need
Full access? Get Clinical Tree
Selecting a Quantitative Research Design
