Nonexperimental designs


Nonexperimental designs

Geri LoBiondo-Wood and Judith Haber


Go to Evolve at for review questions, critiquing exercises, and additional research articles for practice in reviewing and critiquing.

Many phenomena relevant to nursing do not lend themselves to an experimental design. For example, nurses studying cancer-related fatigue may be interested in the amount of fatigue, variations in fatigue, and patient fatigue in response to chemotherapy. The investigator would not design an experimental study that would potentially intensify an aspect of a patient’s fatigue just to study the fatigue experience. Instead, the researcher would examine the factors that contribute to the variability in a patient’s cancer-related fatigue experience using a nonexperimental design. Nonexperimental designs are used in studies in which the researcher wishes to construct a picture of a phenomenon (variable); explore events, people, or situations as they naturally occur; or test relationships and differences among variables. Nonexperimental designs may construct a picture of a phenomenon at one point or over a period of time.

In experimental research the independent variable is actively manipulated; in nonexperimental research it is not. In nonexperimental research the independent variables have naturally occurred, so to speak, and the investigator cannot directly control them by manipulation. In a nonexperimental design the researcher explores relationships or differences among the variables. Even though the researcher does not actively manipulate the variables, the concepts of control and potential sources of bias (see Chapter 8) should be considered as much as possible. Nonexperimental research designs provide Level IV evidence. The strength of evidence provided by nonexperimental designs is not as strong as that for experimental designs because there is a different degree of control within the study; that is, the independent variable is not manipulated, subjects are not randomized, and there is no control group. Yet the information yielded by these types of studies is critical to developing a base of evidence for practice and may represent the best evidence available to answer research or clinical questions.

Researchers are not in agreement on how to classify nonexperimental studies. A continuum of quantitative research design is presented in Figure 10-1. Nonexperimental studies explore the relationships or the differences between variables. This chapter divides nonexperimental designs into survey studies and relationship/difference studies as illustrated in Box 10-1. These categories are somewhat flexible, and other sources may classify nonexperimental studies in a different way. Some studies fall exclusively within one of these categories, whereas many other studies have characteristics of more than one category (Table 10-1). As you read the research literature you will often find that researchers who are conducting a nonexperimental study use several design classifications for one study. This chapter introduces the various types of nonexperimental designs and discusses their advantages and disadvantages, the use of nonexperimental research, the issues of causality, and the critiquing process as it relates to nonexperimental research. The Critical Thinking Decision Path outlines the path to the choice of a nonexperimental design.

Survey studies

The broadest category of nonexperimental designs is the survey study. Survey studies are further classified as descriptive, exploratory, or comparative. Surveys collect detailed descriptions of variables and use the data to justify and assess conditions and practices or to make plans for improving health care practices. You will find that the terms exploratory, descriptive, comparative, and survey are used either alone, interchangeably, or together to describe a study’s design (see Table 10-1).

• Investigators use a survey design to search for accurate information about the characteristics of particular subjects, groups, institutions, or situations or about the frequency of a variable’s occurrence, particularly when little is known about the variable. Box 10-2 provides examples of survey studies.


• Williams and colleagues (2012) conducted a survey of 150 family caregivers to adults with cancer visiting an outpatient chemotherapy center. The purpose of the survey was to determine whether barriers to meditation differ by age and gender among a sample of cancer family caregivers who were chosen because they represent a highly stressed segment of the general population who would likely benefit from meditation practice.

• Moceri and Drevdahl (2012) conducted a survey to investigate emergency nurses’ knowledge and attitudes about pain using the Knowledge and Attitudes Survey Regarding Pain.

• The types of variables in a survey can be classified as opinions, attitudes, or facts.

• Fact variables include attributes of an individual such as gender, income level, political and religious affiliations, ethnicity, occupation, and educational level.

• Surveys provide the basis for further development of programs and interventions.

• Surveys are described as comparative when used to determine differences between variables.

• Survey data can be collected with a questionnaire or an interview (see Chapter 14).

• Surveys have either small or large samples of subjects drawn from defined populations, can be either broad or narrow, and can be made up of people or institutions.

• The data might provide the basis for projecting programmatic needs of groups.

• A survey’s scope and depth depend on the nature of the problem.

• Surveys attempt to relate one variable to another or assess differences between variables, but do not determine causation.

The advantages of surveys are that a great deal of information can be obtained from a large population in a fairly economical manner, and that survey research information can be surprisingly accurate. If a sample is representative of the population (see Chapter 12), even a relatively small number of subjects can provide an accurate picture of the population.

Survey studies have several disadvantages. First, the information obtained in a survey tends to be superficial. The breadth rather than the depth of the information is emphasized. Second, conducting a survey requires a great deal of expertise in various research areas. The survey investigator must know sampling techniques, questionnaire construction, interviewing, and data analysis to produce a reliable and valid study.

Relationship and difference studies

Investigators also try to trace the relationships or differences between variables that can provide a deeper insight into a phenomenon. These studies can be classified as relationship or difference studies. The following types of relationship/difference studies are discussed: correlational studies and developmental studies.

Correlational studies

In a correlational study the relationship between two or more variables is examined. The researcher is:

The direction of the relationship is important (see Chapter 16 for an explanation of the correlation coefficient). For example, in their correlational study, Pinar and colleagues (2012) assessed the relationship between social support and anxiety levels, depression, and quality of life in Turkish women with gynecologic cancer. This study tested multiple variables to assess the relationship and differences among the sample. The researchers concluded that having higher levels of social support were significantly related to lower levels of depression and anxiety, and higher levels of quality of life. Thus the variables were related to (not causal of) outcomes. Each step of this study was consistent with the aims of exploring the relationship among variables.

When reviewing a correlational study, remember what relationship the researcher tested and notice whether the researcher implied a relationship that is consistent with the theoretical framework and hypotheses being tested. Correlational studies offer the following advantages:

The following are disadvantages of correlational studies:

Correlational studies may be further labeled as descriptive correlational or predictive correlational. Given the level of evidence provided by these studies, the ability to generalize the findings has some limitations, but often authors conclude the article with some very thoughtful recommendations for future studies in the specific area. The inability to draw causal statements should not lead you to conclude that a nonexperimental correlational study uses a weak design. In terms of evidence for practice, the researchers, based on the literature review and their findings, frame the utility of the results in light of previous research, and therefore help to establish the “best available” evidence that, combined with clinical expertise, informs clinical decisions regarding the study’s applicability to a specific patient population. A correlational design is a very useful design for clinical research studies because many of the phenomena of clinical interest are beyond the researcher’s ability to manipulate, control, and randomize.

Developmental studies

There are also classifications of nonexperimental designs that use a time perspective. Investigators who use developmental studies are concerned not only with the existing status and the relationship and differences among phenomena at one point in time, but also with changes that result from elapsed time. The following three types of developmental study designs are discussed: cross-sectional, longitudinal, cohort or prospective and retrospective (also labeled as ex post facto or case control). Remember that in the literature, however, studies may be designated by more than one design name. This practice is accepted because many studies have elements of several designs. Table 10-1 provides examples of studies classified with more than one design label.

Cross-sectional studies

A cross-sectional study examines data at one point in time; that is, data collected on only one occasion with the same subjects rather than with the same subjects at several time points. For example, in a study of post-traumatic stress symptoms Melvin and colleagues (2012; see Appendix D) hypothesized that couple functioning, as perceived by each member of the couple, would be negatively associated with post-traumatic stress symptoms (PTSS). This study tested multiple variables (age, gender, military ranks, and marital stress) to assess the relationship and differences among the sample. The researchers concluded that having higher levels of PTSS were associated with lower couple functioning and resilience, thus the variables were related to (not causal of) outcomes. Each step of this study was consistent with the aims of exploring the relationship and differences among variables in a cross-sectional design.

In this study the sample subjects participated on one occasion; that is, data were collected on only one occasion from each subject and represented a cross section of military couples rather than the researchers following a group of military couples over time. The purpose of this study was not to test causality, but to explore the potential relationships between and among variables that can be related to PTSS. Cross-sectional studies can explore relationships and correlations, or differences and comparisons, or both.

Stay updated, free articles. Join our Telegram channel

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

Full access? Get Clinical Tree

Get Clinical Tree app for offline access