A study's design dictates the strength of its causal claims. Learn how to navigate the Oxford CEBM framework, identify publication formats, and match questions to appropriate experimental or observational designs.
Module complete. It is checked off in your course progress.
Step 1 of 4
The first major split in scientific evidence: generating original observations vs synthesizing existing studies.
Primary research gathers brand-new empirical measurements directly. This includes laboratory assays, greenhouse experiments, clinical trials, or field observations.
Secondary research synthesizes existing published literature. This includes narrative literature reviews, systematic reviews, or meta-analyses.
Type of Work
Choose how your project is structured overall. Primary research gathers new observations, while secondary research synthesizes existing literature.
Research Focus
Original research methods differ depending on whether you are studying human subjects, basic biological/physical systems, or documenting a single unique occurrence.
Research Focus
Both systematic reviews and meta-analyses search databases systematically, but they differ in whether they mathematically combine separate outcomes.
Intervention
This is the primary boundary in clinical research: in experimental trials, you assign treatments; in observational studies, exposures happen naturally.
Design/Timeline
Randomization balances known and unknown variables across comparison groups, making it the strongest tool for isolating causality.
Design/Timeline
The direction of measurement in time dictates the strength of your causal claims and the types of biases you must control.
Primary Research
An empirical investigation conducted in a controlled laboratory, greenhouse, field, or computational environment using non-human models (cells, animals, molecules, algorithms, or materials). Generates primary data to test basic scientific theories.
Strengths
Limitations
Example in practice
A student grows Arabidopsis thaliana plants under three different light spectrums in a growth chamber to measure changes in photosynthetic rate and chlorophyll content.
Sources: Grimes, D.A. & Schulz, K.F. (2002). An overview of clinical research. Lancet, 359(9300), 57–61. | Valenzuela, M. et al. (2018). Guidelines for laboratory notebooks. Journal of Biotech, 4(1), 12-15.
Secondary Research
A qualitative overview of the existing literature on a particular topic. It provides a broad narrative summary of current knowledge, trends, and debates, but does not follow a strict, pre-planned database search protocol.
Strengths
Limitations
Example in practice
A student writes a comprehensive review paper summarizing the last 20 years of research on the role of gut microbiome diversity in honeybee colony health.
Sources: Green, B.N. et al. (2006). Writing narrative literature reviews. Journal of Chiropractic Medicine, 5(3), 101-117. | Grant, M.J. & Booth, A. (2009). A typology of reviews. Health Information & Libraries Journal, 26(2), 91-108.
Methodology Development
A paper presenting a new experimental protocol, software tool, mathematical algorithm, or laboratory assay. It focuses on validating that the new method is reliable, accurate, and faster or cheaper than standard techniques.
Strengths
Limitations
Example in practice
A team publishes a detailed protocol for a low-cost, 3D-printable spectrophotometer, demonstrating its accuracy in measuring protein concentration compared to a $5,000 lab model.
Sources: Baker, M. (2016). 1,500 scientists lift the lid on reproducibility. Nature, 533(7604), 452-454. | Nature Protocols Author Guidelines (2025).
Experimental Human Study
Participants are randomly allocated to an intervention group (active treatment) or a control group (placebo or standard care), then followed to compare outcomes. Randomisation balances known and unknown variables to isolate the treatment's effect.
Strengths
Limitations
Example in practice
High school volunteers are randomly assigned to use a specific sleep tracking app (intervention) or a simple paper log (control) for 4 weeks to compare self-reported daytime sleepiness.
Sources: Schulz, K.F. et al. (2010). CONSORT 2010 Statement. BMJ, 340, c332. | Grimes, D.A. & Schulz, K.F. (2002). Lancet, 359, 57–61.
Experimental (Non-randomised)
Tests the effect of an intervention (a new drug, program, teaching style) in human groups, but without random assignment. Participants self-select, are grouped by convenience, or are in pre-existing cohorts.
Strengths
Limitations
Example in practice
A teacher introduces a new hands-on physics simulator in one classroom and compares their test scores at the end of the year to another classroom using the traditional textbook.
Sources: Shadish, W.R., Cook, T.D. & Campbell, D.T. (2002). Experimental and Quasi-Experimental Designs. Houghton Mifflin.
Observational Causal Study
Follows a group of healthy individuals classified by their exposure status forward in time (prospective) to observe who develops the outcome. Excellent for seeing if the exposure preceded the disease.
Strengths
Limitations
Example in practice
A researcher follows a group of 300 athletes—some who consistently wear custom orthotics and some who do not—over a 3-year period to observe the occurrence of stress fractures.
Sources: Oxford Centre for Evidence-Based Medicine (2021). Study Designs. | Lash, T.L. et al. (2021). Modern Epidemiology (4th ed.). Wolters Kluwer.
Observational Causal Study
Starts with the outcome: individuals who already have the condition (cases) and matched individuals without it (controls) are recruited. Researchers look backward in time (retrospective) to compare exposures.
Strengths
Limitations
Example in practice
A researcher compares 50 pediatric patients diagnosed with a rare childhood asthma subtype (cases) to 50 healthy children (controls) to examine historical household allergen exposure.
Sources: Schlesselman, J.J. (1982). Case-Control Studies. Oxford University Press. | Grimes, D.A. & Schulz, K.F. (2002). Lancet, 359, 57–61.
Observational descriptive / snapshot
Measures both exposure levels and outcomes in a study population at a single moment in time (a 'snapshot'). Can check for correlations but cannot prove which factor came first.
Strengths
Limitations
Example in practice
A survey measures the daily caffeine intake and current anxiety level of 100 college students on a Tuesday afternoon during final exams.
Sources: Oxford Centre for Evidence-Based Medicine (2021). | Rezigalla, A.A. (2020). Cureus, 12(1), e6692.
Descriptive Clinical Study
Detailed description of a single patient (case report) or a small cluster of patients (case series) presenting with an unusual symptom, rare disease, unexpected treatment response, or side effect.
Strengths
Limitations
Example in practice
A physician publishes a report detailing the clinical presentation, diagnosis, and successful treatment of a patient who developed a rare autoimmune reaction after taking a common antibiotic.
Sources: Rezigalla, A.A. (2020). Observational Study Designs. Cureus, 12(1), e6692.
Secondary Research (Synthesis)
A structured, exhaustive synthesis of all primary research papers answering a specific question. It uses a pre-planned, transparent, and reproducible search protocol to scan databases, screen studies, and grade evidence quality.
Strengths
Limitations
Example in practice
A review team screens 1,200 abstracts from PubMed and Scopus to synthesize all published clinical trials on the effectiveness of mindfulness meditation for reducing blood pressure in adults.
Sources: Liberati, A. et al. (2009). The PRISMA Statement. PLOS Medicine, 6(7), e1000097. | Higgins, J.P.T. et al. (Eds.) (2022). Cochrane Handbook for Systematic Reviews.
Secondary Research (Quantitative Synthesis)
A specific type of systematic review that uses statistical techniques to combine and pool the numerical data from multiple separate primary studies, generating a single, precise, and weighted mathematical effect size.
Strengths
Limitations
Example in practice
A researcher extracts the odds ratios of 15 independent cohort studies and statistically pools them to calculate the overall relative risk of stroke in patients with untreated sleep apnea.
Sources: Borenstein, M. et al. (2009). Introduction to Meta-Analysis. Wiley. | Higgins, J.P.T. et al. (Eds.) (2022).