Glossary of Terms Related to Research Design
A pre-post investigation of a discrete procedure, experience or event that is not managed by the researcher. Data are collected at baseline and one or more times after the procedure, experience or event.
Case Control Study
A study which involves identifying patients who have the outcome of interest (cases) and matching them with individuals who have similar characteristics, but do not have the outcome of interest (controls), and then looking back to see if these two groups differed with regard to the exposure of interest (i.e., the hypothesized causal or contributing factors).
Case Study or Case Series
A descriptive study of one (case study or case report) or a series of patients (case series) defined by eligibility criteria, and where the unfolding course of events (disease progression, therapies, outcomes, etc.) is described in detail. The study researchers do not manipulate interventions. This study design is used to provide a detailed description of an uncommon disease or condition, a unique situation, or the introduction of a new technique.
Cluster Randomized Trial
A special type of a randomized controlled trial (RCT) where groups of individuals (e.g., clinic sites, classrooms, communities), rather than independent individuals, are randomized to the intervention alternatives.
A study that involves the identification of a group (cohort) of individuals with specific characteristics in common and follows them over time to gather data about exposure to factors and the development of the outcome of interest. Comparison groups can be defined at the beginning or created later using data from the study (e.g., age group, smokers/non-smokers, amount of a specific food group consumed). Prospective cohort studies enroll individuals and then collect data at many intervals. Retrospective cohort studies use an existing longitudinal data set to look back for a temporal relationship between exposure factors and outcome development. In the medical field, many studies labeled a “population-based clinical study” could be classified as retrospective cohort studies.
Cost Benefit Analysis or Cost Effectiveness Analysis
An analysis that assesses the cost of an intervention in relation to the magnitude of outcome achieved. In cost benefit analysis, the inputs (i.e., intervention alternatives) and the resulting outcomes are quantified and expressed in monetary terms. In cost effectiveness analysis, inputs (i.e., intervention alternatives) are expressed in monetary terms but the outcomes are expressed in a standard unit, such as quality adjusted life years (QALY) or hospitalizations avoided. These are considered a synthesis of primary studies when data from multiple studies are used to derive estimates of inputs and outcomes.
Crossover Study Design
A study where two or more experimental therapies are administered, one after the other, in a specified or a random sequence, to the same group of patients. Usually there is a washout (no treatment) period between therapies. Individuals serve as their own controls. A crossover study is a special type of a randomized or non-randomized trial.
A study where exposure factors (e.g., individual or environmental risk factor, nutrition education) and outcomes (e.g., disease occurrence, eating behavior) are observed or measured at one point in time in a sample from the population of interest, usually by survey or interview. In this design, a researcher examines the association among factors and outcomes using a statistical test for association, but cannot infer cause and effect.
Descriptive studies, as a research category, use a variety of methods to observe existing natural or man-made phenomena without influencing it (no researcher intervention). Data are gathered, organized and analyzed to depict and describe “what is”. Descriptive studies can be quantitative and/or qualitative and provide an in-depth look at processes, characteristics and patterns. Descriptive studies can result in a theory or framework, but they do not try to determine cause and effect.
Diagnostic, Validity or Reliability Study
Types of studies that are designed to determine the sensitivity and specificity of diagnostic and assessment methods and the accuracy and/or consistency of tests or tools used to measure variables and concepts.
Epidemiological studies, as a research category, are analytical studies of the determinants of health and illness in specific populations. Studies are designed to determine the relationship among exposure factors (which can be risk factors or protective factors) and outcomes. Epidemiologic studies are observational; the researcher does not manage any intervention. The most common epidemiological study designs are case-control, cohort, and cross-sectional studies.
Intention to Treat Analysis
A method of analysis for intervention trials in which all patients originally assigned to a treatment group are included in the analysis for that group, regardless of whether or not they completed or received that treatment.
A general term that indicates data are collected from the same subjects at several points over time. It is not a specific study design.
Magnitude of Effect
Refers to how much change can be attributed to the treatment or intervention in a particular study.
A systematic, quantitative method that combines the results of all relevant studies to produce an overall estimate. A meta-analysis can be part of a systematic review, but not all systematic reviews include meta-analysis.
A summary report of the state of knowledge on a particular topic. Narrative reviews are less rigorous than systematic reviews in that search methods, study inclusion criteria, and quality of the studies are often not reported.
A type of intervention trial where only one group is used (there is no comparison group); but the studied intervention is defined and managed by the researcher.
Non-Randomized Controlled Trial
A study where subjects are assigned to intervention (protocol, method or treatment) alternatives by a method that is not random. The researcher does define and manage the alternatives, which could be treatment and control or two or more different interventions.
Observational studies include a wide range of studies in which the course of events is studied as it unfolds. The researcher does not intervene. Changes or differences that occur between groups are used to draw inferences about the association of variables and the relationships between possible causal factors and outcomes.
Any event, circumstance, or experience that is apparent to the senses and that can be scientifically described or appraised.
Prospective Cohort Study
See Cohort Study.
Randomized Controlled Trial (RCT)
Individuals meeting eligibility requirements are randomly assigned into an experimental group or a control group. The experimental intervention (protocol, method or treatment) and its alternative(s) are clearly defined and their implementation is closely managed by the researcher.
Retrospective Cohort Study
See Cohort Study.
See Narrative Review or Systematic Review.
A summary of the scientific literature on a specific topic or question that uses explicit methods to conduct a comprehensive literature search and identify relevant studies, critically appraise the quality of each study, and summarize the body of literature or evidence to answer the question.
A study collecting data from the same subjects at a series of points over time during which a discrete preventive or therapeutic procedure, life experience, or event takes place. Data are collected prior to, and after (and sometimes during) the event in order to reach conclusions about its effect. Some studies labeled as “longitudinal” are time series studies.
An experimental or quasi-experimental study to determine the effect of an intervention.
A study in which the same or similar data about exposures and outcomes are collected from the same population many times, but each time a different sample is used. A trend study is like a series of cross-sectional studies. An example is NHANES.
- Study Designs
Reserarch Designs and Hierarchy
Which type of study is preferred?
The four most common types of evidence analysis questions are: diagnosis, treatment, prognosis and etiology.The type of question you are trying to answer determines the best research design to seek.
For instance, a randomized controlled trial (RCT) would be the most appropriate type of study to anwer a question about therapy or treatment. This hierarchy is often shown graphically as a pyramid with expert opinions at the bottom of the pyramid and randomized controlled trials (RCTs) at the top. However, a RCT would not be the the strongest research design to answer a questoin about prognosis. The highest level of evidence for prognosis is a cohort study. Always look for the strongest evidence you can find to answer your type of question.
In some situations the eligibility of research article depends on the research design used. These resources will help identify the study design: