Look past the headline
Explore what different study designs can show, then change example numbers to see why a percentage needs context.
How do you read cannabis research?
Start with the research question, then check who or what was studied, the comparison, the outcome, and the limitations. A headline alone cannot tell you whether a study observed a relationship, tested an intervention, or explored a mechanism.
A result should stay connected to its study design and conditions. Use the tools below to practice asking useful questions before drawing a conclusion.
All examples and numbers on this page are fictional teaching aids. They are not findings from actual cannabis studies, product recommendations, or guidance on treatment or dosing.
Choose a study type
Select a design to see its purpose, its limits, and a question to ask when reading it.
Patterns in people, without assigning an intervention
Fictional example: a survey compares reported cannabis use with sleep scores.
Find associations and generate hypotheses. These studies can also examine exposures that researchers cannot ethically assign.
A single association does not establish that cannabis caused the outcome. Other factors, the timing of events, and how participants were selected can affect interpretation.
Ask: Did the analysis address plausible alternative explanations? Source
Exploring biology before claiming a human benefit
Fictional example: a cannabinoid changes a biological marker in mice.
Investigate mechanisms and identify questions worth studying further.
An animal or cell result does not demonstrate a treatment benefit in people or establish a human dose.
Ask: Has the finding been investigated in humans, using an appropriate outcome? Source
Comparing assigned interventions
Fictional example: adults are randomly assigned a defined formulation or a comparison intervention.
Help test whether an intervention causes an outcome under the conditions studied.
Randomization is not a guarantee of flawless evidence. Study size, blinding, missing data, comparison choice, and follow-up still matter. Findings may not apply to other preparations or populations.
Ask: What was the between-group difference, and how uncertain was it? Source
Checking whether a larger study is feasible
Fictional example: 12 volunteers help researchers test recruitment, visit schedules, and measurement procedures.
Assess whether the methods are practical and acceptable before a larger study.
A small feasibility pilot is not designed to establish effectiveness. Seeing no safety problem in a small pilot does not establish safety.
Ask: Was this study designed to test benefit, or to test the study process? Source
One difference, two ways to describe it
In a fictional comparison, 20% of Group A and 10% of Group B report an outcome. That is a difference of 10 percentage points. It is also a 50% reduction relative to the Group A rate.
Change the rates below. These are percentages of people reporting an unspecified outcome. A lower number is not automatically better.
Use values from 0 to 100, in increments of 0.1. Calculations update as you type.
Enter a number from 0 to 100 in both fields, using at most one decimal place.
Check what participants actually received
Cannabis research can examine whole plant material, extracts, individual compounds, or defined medicines. Keep the compound or preparation, route of administration, population, and outcome attached to the finding. NCCIH’s cannabis overview distinguishes plant material and constituents from specific cannabinoid medicines. [3]
For example, a fictional trial of a purified formulation would not by itself establish the same effects for every type of cannabis flower. That is a question about whether the findings apply beyond the tested intervention.
A short checklist for the next headline
- Open the original source. Find the study title, date, methods, and results.
- Identify the design. Was it a survey, an animal or cell experiment, or a human trial?
- Find the comparison. Identify what was compared, how participants were selected or assigned, and what was measured.
- Read the numbers in context. Look for group sizes, baseline rates, absolute differences, uncertainty, and follow-up length.
- Keep unanswered questions visible. Describe what the study supports and what remains unresolved.
For a different kind of evidence-reading practice, try the Cannabis Lab Report Quiz. A lab report measures an identified sample; a health study investigates questions about biology or outcomes.
Common questions
Does correlation mean a study is worthless?
No. Observational research can identify patterns and help investigate risks. The issue is treating one association as sufficient proof of causation. [1]
Does a small study prove there are no side effects?
No. Uncommon events may not appear in a small sample, and short follow-up does not describe long-term outcomes. [4]
Why can a relative change look so large?
It is divided by the starting rate. Moving from 2% to 1% is only a 1 percentage-point difference but a 50% relative reduction. Always ask for the baseline and absolute difference.
Sources and scope
- NIH: Understanding Clinical Studies. Study designs and cause-and-effect evidence.
- NCCIH: Research Framework. Basic, translational, and clinical research.
- NCCIH: Cannabis and Cannabinoids, What You Need To Know. Context on cannabis constituents, preparations, and research.
- NCCIH: Pilot Studies, Common Uses and Misuses. Feasibility studies and the limits of small samples.
The calculator uses percentage-point difference = Group B rate minus Group A rate, and relative change = that difference divided by Group A rate, multiplied by 100. A zero baseline makes relative change undefined. Displayed results are rounded to two decimal places.
Chris Garcia, Owner and CEO
September 10, 2026
How this page is prepared: Plenny uses linked research and other relevant sources for educational explanations, and product details and batch COAs where relevant. General strain, flavor, and effect descriptions are not guarantees.
Commercial transparency: Plenny sells products discussed on this website. Educational pages are not medical or legal advice, and content is not medically reviewed unless a named licensed reviewer is shown.