Verifiedintermediate
Before you survey anyone
Consent, assent, minors, and the approval that must come before recruitment — not after.
- consent
- irb
- surveys
- +1
Human Participants8 min
Resources
Curated guides, templates, and reading — the things worth knowing before you need them.
21 resources
Verifiedintermediate
Choosing a statistical test
A decision path from your design to the right test, and what to do when the assumptions fail.
- tests
- t-test
- anova
- +1
Statistics8 min
Verifiedintermediate
Common pitfalls in machine-learning projects
scikit-learn's own guide to data leakage, improper splits, and the mistakes that inflate accuracy.
- leakage
- validation
- scikit-learn
Machine Learning20 min
Verifiedintermediate
Designing an experiment that survives review
Variables, controls, replication, randomisation, and blinding — what each one is protecting you from.
- design
- controls
- replication
- +1
Experimental Design9 min
Verifiedintermediate
Official ISEF international rules
The current rules and forms. Always download from here rather than reusing a copy from last year — numbering and requirements change.
- rules
- forms
- isef
- +1
Vertebrate Animals45 min
Verifiedintermediate
Reading a paper without reading all of it
The three-pass method: how to decide in ten minutes whether a paper is worth an hour.
- reading
- triage
- notes
Literature Reviews7 min
Verifiedintermediate
Research integrity for student projects
Attribution, data handling, and the grey areas that get projects disqualified.
- integrity
- attribution
- honesty
Research Ethics7 min
Verifiedintermediate
Structuring an analysis you can defend
A folder layout and a set of habits that keep a computational project reproducible.
- reproducibility
- code
- structure
Programming7 min
Verifiedintermediate
The 250-word abstract, sentence by sentence
A seven-sentence skeleton that fits the usual word limit and covers everything judges look for.
- abstract
- structure
- word-limit
Abstract Writing6 min
Verifiedintermediate
The questions judges actually ask
Twelve recurring interview questions and what a strong answer contains.
- judging
- interview
- questions
Verifiedintermediate
What separates a finalist project from a good one
Patterns that recur in projects that place, drawn from judging rubrics rather than from any single project.
- judging
- rubrics
- quality
Example Projects8 min
Verifiedintermediate
Choosing the right control
Negative, positive, vehicle, and sham controls — and how to tell which your experiment needs.
- controls
- baseline
- confounds
Control Groups6 min
Verifiedintermediate
Growing microbes safely in a school setting
Sterile technique, sealed plates, and a disposal plan you can point a reviewer at.
- sterile-technique
- culture
- disposal
Biological Agents6 min
Verifiedintermediate
Hedging: saying exactly how sure you are
The vocabulary of scientific caution, and why over-claiming costs more than it gains.
- claims
- hedging
- precision
Scientific Writing6 min
Verifiedintermediate
Null and alternative hypotheses, plainly
What the null hypothesis is actually for, and why "I rejected the null" is not the same as "I proved my idea".
- statistics
- null-hypothesis
- inference
Hypothesis Development15 min
Verifiedintermediate
Operational definitions
Turning "healthier", "faster", and "better" into something two people would measure the same way.
- measurement
- definitions
- reproducibility
Verifiedintermediate
Reading a safety data sheet
Which of the sixteen SDS sections actually change your protocol, and what to extract from each.
- sds
- ppe
- risk-assessment
Hazardous Chemicals6 min
Verifiedintermediate
Risk assessment for build projects
A five-column method for identifying and controlling hazards in mechanical, electrical, and thermal builds.
- risk-assessment
- electrical
- mechanical
Engineering Safety8 min
Verifiedintermediate
Structuring a literature review
Why chronological reviews read badly, and what to organise by instead.
- writing
- structure
- argument
Literature Reviews6 min
Verifiedintermediate
Writing survey questions that do not lead
Wording, scales, and order effects — the bias you build in before anyone answers.
- surveys
- bias
- questionnaire
Human Participants7 min
Verifiedintermediate
Your model needs a baseline
Why 94% accuracy can be a terrible result, and what to compare against instead.
- baseline
- evaluation
- metrics
Machine Learning7 min