Purchases are not available yet. You can browse the site and sign in.
When someone says "studies show," the natural question is which kind of study? A friend's report, a survey of thousands, a single clinical trial, and a summary of fifty trials are all called "a study" in casual speech, yet they carry wildly different weight. Researchers sort them into a rough hierarchy — often pictured as a pyramid — from weaker evidence at the bottom to stronger at the top, specifically for cause-and-effect questions like "does this actually do anything?"
From weaker to stronger:
Anecdote and case reports — one person's experience, or a clinician's note on a single patient. Vivid, hypothesis-generating, but it cannot separate the substance from coincidence, mood, or expectation.
Observational and survey studies — large groups answering questionnaires or tracked over time. They reveal patterns and correlations, but the people often chose what they did, so cause and effect stay tangled.
Randomised controlled trials (RCTs) — people assigned by chance to a real or comparison condition. This design is built to isolate cause, which is why even a single small RCT can outrank a huge survey.
Systematic reviews and meta-analyses — structured syntheses that gather many trials and pool their results, smoothing out the noise of any one study.
This is the part that surprises people. A survey of 5,000 microdosers who report better focus feels powerful — that's a lot of people. But everyone in it chose to microdose. Maybe the kind of person who tries it is already curious, motivated, or optimistic. The survey cannot tell you whether the substance caused the focus or whether focused people were drawn to the substance.
For cause-and-effect questions, design beats headcount. A small randomised trial that compares against a control can answer 'better than what?' — a question a giant survey cannot touch.
A 40-person RCT that randomly assigns people, so the groups start out comparable, is built to answer that. Numbers still matter — we'll spend a later lesson on sample size — but a clever design with few people often tells you more about causation than a large study with none.
Before you weigh a paper, work out what you're actually holding. Check the methods section — it tells you instantly.
Answer every question, then submit to see how you did. You can retake it as often as you like — your score is just for you.
Members can compare notes on this lesson below.
Lesson discussion is a member perk. Become a member to comment.
Be the first to comment.
A primary study collected new data: it has participants, a procedure, and fresh results. Its methods describe what the researchers did to people or samples.
A review or meta-analysis collected no new data; it gathers and synthesises existing studies. Its methods describe how papers were searched for and selected.
An opinion or commentary piece has no real methods at all — it's an argument, an editorial, a perspective. Useful for ideas, but it is not evidence of an effect.
The trap is that all three can sound equally authoritative in a quote.
Being higher on the pyramid is not a free pass. Two failures hide there:
Cherry-picking in narrative reviews. A narrative review has no systematic search rules, so an author can quietly include the studies that support their view and skip the ones that don't. It reads like a balanced overview while being a curated case for one side.
Garbage in, garbage out. A meta-analysis is only as good as the trials it pools. Combine twenty tiny, unblinded, poorly designed studies and you get a precise-looking average built on sand. The respectable label "meta-analysis" can sit on top of weak ingredients.
Microdosing is a sharp case study. A great deal of what circulates — testimonials, survey write-ups, enthusiastic blog summaries — sits near the bottom of the pyramid, even when it's described as "the research." Knowing the hierarchy lets you place each claim where it belongs without dismissing it or swallowing it whole.
None of this tells you what to do with your own life. It tells you how much weight a given page of "evidence" can actually bear.
Ask first what kind of thing you're holding. The label on the box rarely matches the contents.