What are Analytical Study Designs?

A study design is critical to your research study because it determines exactly how you will collect and analyze your data. If your study aims to study the relationship between two variables, then an analytical study design is the right choice.

But how do you know which type of analytical study design is best for your specific research question? It's necessary to have a clear plan before you begin data collection. Lots of researchers, sadly, speed through this or don't do it at all.

Analytical study designs can be experimental or observational and each type has its own features. In this article, you'll learn the main types of designs and how to figure out which one you'll need for your study.

When are analytical study designs used?

A study design is a systematic plan, developed so you can carry out your research study effectively and efficiently. Having a design is important because it will determine the right methodologies for your study. Using the right study design makes your results more credible, valid, and coherent.

Descriptive vs. analytical studies

Study designs can be broadly divided into either descriptive or analytical.

Descriptive studies describe characteristics such as patterns or trends. They answer the questions of what, who, where, and when, and they generate hypotheses. They include case reports and qualitative studies.

Analytical study designs quantify a relationship between different variables. They answer the questions of why and how. They're used to test hypotheses and make predictions.

Experimental and observational

Analytical study designs can be either experimental or observational. In experimental studies, researchers manipulate something in a population of interest and examine its effects. These designs are used to establish a causal link between two variables.

In observational studies, in contrast, researchers observe the effects of a treatment or intervention without manipulating anything. Observational studies are most often used to study larger patterns over longer periods.

Experimental study designs

Experimental study designs are when a researcher introduces a change in one group and not in another. Typically, these are used when researchers are interested in the effects of this change on some outcome. It's important to try to ensure that both groups are equivalent at baseline to make sure that any differences that arise are from any introduced change.

In one study, Reiner and colleagues studied the effects of a mindfulness intervention on pain perception. The researchers randomly assigned participants into an experimental group that received a mindfulness training program for two weeks. The rest of the participants were placed in a control group that did not receive the intervention.

Experimental studies help us establish causality. This is critical in science because we want to know whether one variable leads to a change, or causes another. Establishing causality leads to higher internal validity and makes results reproducible.

Experimental designs include randomized control trials (RCTs), nonrandomized control trials (non-RCTs), and crossover designs. Read on to learn the differences.

Randomized control trials

In an RCT, one group of individuals receives an intervention or a treatment, while another does not. It's then possible to investigate what happens to the participants in each group.

Another important feature of RCTs is that participants are randomly assigned to study groups. This helps to limit certain biases and retain better control. Randomization also lets researchers pinpoint any differences in outcomes to the intervention received during the trial. RTCs are considered the gold standard in biomedical research and are considered to provide the best kind of evidence.

For example, one RCT looked at whether an exercise intervention impacts depression. Researchers randomly placed patients with depressive symptoms into intervention groups containing different types of exercise (i.e., light, moderate, or strong). Another group received usual medications or no exercise interventions.

Results showed that after the 12-week trial, patients in all exercise groups had decreased depression levels compared to the control group. This means that by using an RCT design, researchers can now safely assume that the exercise variable has a positive impact on depression.

However, RCTs are not without drawbacks. In the example above, we don't know if exercise still has a positive impact on depression in the long term. This is because it's not feasible to keep people under these controlled settings for a long time.

Advantages of RCTs

• It is possible to infer causality
• Everything is properly controlled, so very little is left to chance or bias
• Can be certain that any difference is coming from the intervention

Disadvantages of RCTs

• Expensive and can be time-consuming
• Can take years for results to be available
• Cannot be done for certain types of questions due to ethical reasons, such as asking participants to undergo harmful treatment
• Limited in how many participants researchers can adequately manage in one study or trial
• Not feasible for people to live under controlled conditions for a long time

Nonrandomized controlled trials

Nonrandomized controlled trials are a type of nonrandomized controlled studies (NRS) where the allocation of participants to intervention groups is not done randomly. Here, researchers purposely assign some participants to one group and others to another group based on certain features. Alternatively, participants can sometimes also decide which group they want to be in.

For example, in one study, clinicians were interested in the impact of stroke recovery after being in an enriched versus non-enriched hospital environment. Patients were selected for the trial if they fulfilled certain requirements common to stroke recovery. Then, the intervention group was given access to an enriched environment (i.e. internet access, reading, going outside), and another group was not. Results showed that the enriched group performed better on cognitive tasks.

NRS are useful in medical research because they help study phenomena that would be difficult to measure with an RCT. However, one of their major drawbacks is that we cannot be sure if the intervention leads to the outcome. In the above example, we can't say for certain whether those patients improved after stroke because they were in the enriched environment or whether there were other variables at play.

Advantages of NRS's

• Good option when randomized control trials are not feasible
• More flexible than RCTs

Disadvantages of NRS's

• Can't be sure if the groups have underlying differences
• Introduces risk of bias and confounds

Crossover study

In a crossover design, each participant receives a sequence of different treatments. Crossover designs can be applied to RCTs, in which each participant is randomly assigned to different study groups.

For example, one study looked at the effects of replacing butter with margarine on lipoproteins levels in individuals with cholesterol. Patients were randomly assigned to a 6-week butter diet, followed by a 6-week margarine diet. In between both diets, participants ate a normal diet for 5 weeks.

These designs are helpful because they reduce bias. In the example above, each participant completed both interventions, making them serve as their own control. However, we don't know if eating butter or margarine first leads to certain results in some subjects.

Advantages of crossover studies

• Each participant serves as their own control, reducing confounding variables
• Require fewer participants, so they have better statistical power

Disadvantages of crossover studies

• Susceptible to order effects, meaning the order in which a treatment was given may have an effect
• Carry-over effects between treatments

Observational studies

In observational studies, researchers watch (observe) the effects of a treatment or intervention without trying to change anything in the population. Observational studies help us establish broad trends and patterns in large-scale datasets or populations. They are also a great alternative when an experimental study is not an option.

Unlike experimental research, observational studies do not help us establish causality. This is because researchers do not actively control any variables. Rather, they investigate statistical relationships between them. Often this is done using a correlational approach.

For example, researchers would like to examine the effects of daily fiber intake on bone density. They conduct a large-scale survey of thousands of individuals to examine correlations of fiber intake with different health measures.

The main observational studies are case-control, cohort, and cross-sectional. Let's take a closer look at each one below.

Case-control study

A case-control is a type of observational design in which researchers identify individuals with an existing health situation (cases) and a similar group without the health issue (controls). The cases and the controls are then compared based on some measurements.

Frequently, data collection in a case-control study is retroactive (i.e., backwards in time). This is because participants have already been exposed to the event in question. Additionally, researchers must go through records and patient files to obtain the records for this study design.

For example, a group of researchers examined whether using sleeping pills puts people at risk of Alzheimer's disease. They selected 1976 individuals that received a dementia diagnosis (“cases”) with 7184 other individuals (“controls”). Cases and controls were matched on specific measures such as sex and age. Patient data was consulted to find out how much sleeping pills were consumed over the course of a certain time.

Case-control is ideal for situations where cases are easy to pick out and compare. For instance, in studying rare diseases or outbreaks.

Advantages of case-control studies

• Feasible for rare diseases
• Cheaper and easier to do than an RCT

Disadvantages of case-control studies

• Relies on patient records, which could be lost or damaged
• Potential recall and selection bias

Cohort study (longitudinal)

A cohort is a group of people who are linked in some way. For instance, a birth year cohort is all people born in a specific year. In cohort studies, researchers compare what happens to individuals in the cohort that have been exposed to some variable compared with those that haven't on different variables. They're also called longitudinal studies.

The cohort is then repeatedly assessed on variables of interest over a period of time. There is no set amount of time required for cohort studies. They can range from a few weeks to many years.

Cohort studies can be prospective. In this case, individuals are followed for some time into the future. They can also be retrospective, where data is collected on a cohort from records.

One of the longest cohort studies today is The Harvard Study of Adult Development. This cohort study has been tracking various health outcomes of 268 Harvard graduates and 456 poor individuals in Boston from 1939 to 2014. Physical screenings, blood samples, brain scans and surveys were collected on this cohort for over 70 years. This study has produced a wealth of knowledge on outcomes throughout life.

A cohort study design is a good option when you have a specific group of people you want to study over time. However, a major drawback is that they take a long time and lack control.

Advantages of cohort studies

• Ethically safe
• Allows you to study multiple outcome variables
• Establish trends and patterns

Disadvantages of cohort studies

• Time consuming and expensive
• Can take many years for results to be revealed
• Too many variables to manage
• Depending on length of study, can have many changes in research personnel

Cross-sectional study

Cross-sectional studies are also known as prevalence studies. They look at the relationship of specific variables in a population in one given time. In cross-sectional studies, the researcher does not try to manipulate any of the variables, just study them using statistical analyses. Cross-sectional studies are also called snapshots of a certain variable or time.

For example, researchers wanted to determine the prevalence of inappropriate antibiotic use to study the growing concern about antibiotic resistance. Participants completed a self-administered questionnaire assessing their knowledge and attitude toward antibiotic use. Then, researchers performed statistical analyses on their responses to determine the relationship between the variables.

Cross-sectional study designs are ideal when gathering initial data on a research question. This data can then be analyzed again later. By knowing the public's general attitudes towards antibiotics, this information can then be relayed to physicians or public health authorities. However, it's often difficult to determine how long these results stay true for.

Advantages of cross-sectional studies

• Fast and inexpensive
• Ethically safe
• Provides a great deal of information for a given time point
• Leaves room for secondary analysis

Disadvantages of cross-sectional studies

• Requires a large sample to be accurate
• Not clear how long results remain true for
• Do not provide information on causality
• Cannot be used to establish long-term trends because data is only for a given time

So, how about your next study?

Whether it's an RCT, a case-control, or even a qualitative study, AJE has services to help you at every step of the publication process. Get expert guidance and publish your work for the world to see.