Study Design Terms
Baseline Data that describes the condition of the population at the beginning of the study. Used for comparison with conditions after the study. For example, researchers might take baseline readings of cholesterol levels before testing a cholesterol-lowering medication and compare those baseline numbers with the levels at the end of the study.

Blinding Keeping information about the study secret. Studies can be double blinded, where both staff and participants don’t know which group the subjects are in, or single blinded where only one side does not know.

Clinical Trial Study done to test a particular medical treatment on human subjects using an experimental approach.

Cohort The clearly defined group being studied. Often a cohort study will look at a group over a long period of time to see what diseases might develop. An example of this would be a study following women who were diagnosed with gestational diabetes for 30 years to see if they develop Type 2 diabetes at a higher rate than the general population.

Control Group In a clinical trial, the group that does NOT receive the intervention being tested. They may receive no treatment, a placebo, or “usual care”, depending on the setup. Some studies may use existing data on the general population as a control group.

Crossover Design A variation on experimental design where the control and experimental groups switch halfway through the study. Essentially the study is run twice, each group having a turn as the experimental group and the control group.

Crossover Effect The amount of people in the experimental and control groups who end up not receiving the intervention they were assigned. High crossover can negate the findings of a study. For example, if a large portion of the women randomized to walk during labor instead chose to stay in bed, and a large portion of the women randomized to stay in bed chose to walk, this would make it hard to show any significant difference in the two groups.

Experimental Group In a clinical trial, the group that receives the intervention being studied.

Hawthorne Effect Well-documented idea that simply knowing they are in a study or being observed changes people’s behavior. Sometimes called “halo effect” as well.

Longitudinal Study Study that follows participants for a length of time to look at long-term effects. Difficult and expensive to maintain contact, and researchers must start with a larger sample size to have sufficient numbers by the end of the study.

Meta-Analysis Combining the data from several studies together and analyzing the results together. Sometimes tricky as studies collect different data and use different definitions and standards.

Methodology The way the study was carried out. Should be spelled out in the article. Readers can read the methods and decide if they were appropriately designed.

Pilot Study A small study done to test the research design and see if it is workable. Usually done before undertaking a larger study, but occasionally results of pilot studies end up in the news.

Population The group of people being studied. It could be very broad, like “all residents of Utah” or very narrow, like “married Hispanic women over 40 in the third trimester of their first pregnancy”. Should be well defined in the study.

Prospective Study A study set up before data collection begins. The researcher can choose what data to collect, making it easier to study new techniques, information not routinely collected, etc. Possible pitfall is bias on the part of researchers.

Qualitative Research Studies that look at experiences and processes. Methods might include surveys, interviews, observations, etc. A good example of this is the “Listening to Mothers” report published by Childbirth Connection.

Quantitative Research Studies that look at hard numbers and aim for accuracy, control and application to the general population. Clinical studies and randomized controlled trials are examples of quantitative research.

Quasi-experimental Studies with an experimental group and a control group, but where there is no random assignment to groups. Works in situations where random assignment may not be ethical, but results are not as strong as in a randomized controlled trial. An example of where randomization might not be ethical is a study comparing cesarean birth and vaginal birth. It would not be ethical to randomly assign a woman to have a cesarean with no medical reason, so a quasi-experimental design would have to do.

Randomized Controlled Trial Sometimes called the “gold standard” of research studies, and often abbreviated “RCT.” Participants are randomly assigned to either an experimental group (which receives the intervention tested) or a control group (which receives either no treatment, a placebo, or the “usual care”) Researchers can then compare the two groups.

Retrospective Study A study done on data that already exists when the researcher sets up the study. Often done with medical records, birth certificate data, etc. Can reduce bias in data collection but is limited to existing data.

Subject A person participating in the research study.

Systemic Review A form of meta-analysis that reviews all studies on a certain topic and uses predetermined criteria to evaluate them. Studies that meet the criteria are included in the meta-analysis and are sometimes weighted based on quality of the study methodology.

Variable The items of interest to a researcher. The independent variable is the thing the researcher is manipulating. The dependant variable is the result. One example might be a study looking at types of induction. The type of induction would be the independent variable, and dependant variables would be things like length of labor, cesarean rates, fetal distress rates, etc.