Evaluating Evidence: Spotting Weak Studies With AI Help
Judge research quality with structured appraisal questions, using AI to help but not to decide.
A taste of a lesson
A headline says a food 'increases cancer risk by 20%'. The study is real. Should I be worried?
First, translate it into absolute numbers. A 20% relative increase on a baseline of 5 in 1,000 people becomes 6 in 1,000, which is one extra case per thousand. Then check the design: if it is observational, confounding is likely, since people who eat that food may differ in other ways. Look for the confidence interval and whether other studies agree. Ask the assistant to quote the baseline risk and the design from the paper, then verify. What baseline risk does the study report?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Explain what each major study design can and cannot show
- Identify common biases, small sample effects and publication bias
- Interpret relative and absolute risk, effect sizes and confidence intervals
- Apply a design specific appraisal checklist with AI help and verify every answer
Lesson plan
- 1 Designs and their questions Know what each study design can show and where it falls short. Start
- 2 Sources of bias Recognise how studies can be misled by design or conduct. Start
- 3 Numbers that mislead Read effects and risks accurately. Start
- 4 The wider literature Understand how publication practices distort fields. Start
- 5 Appraisal with AI support Use a checklist and an assistant together on a real paper. Start
- 6 Proportionate conclusions State what the evidence supports and with what confidence. Start
Try asking
About this tutor
For postgraduates, professionals who rely on research, and serious self learners who need to judge how much a study should count. You learn the main study designs and what each can show, common sources of bias, the difference between relative and absolute risk, why small or unreplicated studies mislead, how publication bias distorts a field, and what preregistration and conflicts of interest tell you. You practise using published critical appraisal checklists, with an assistant helping you work through them and explain concepts, while you verify every answer against the paper. Education only: clinical or policy decisions belong with qualified professionals.
Reviews
Students can review a tutor after a paid lesson. Nobody has yet.
About the teacher
Research with AI assistants done carefully: finding, reading, checking and citing
9 tutors 379 lessons taught Sample
I teach research skills for students, postgraduates and curious adults who want to use AI assistants without fooling themselves. My background is in research support and information literacy, so I spend a lot of time on the unglamorous parts: where a claim came from, whether a citation exists, and what a paper actually says. I am enthusiastic about what assistants...
See Delphine's profile and tutorsMore like this
Other tutors on the same or nearby topics.