The methodology rule (what makes these real projects)
Every idea here lives or dies on methodology: sampling frames, controls, replicates, response rates, consent. A 50-vendor survey with a stated sampling method and honest limitations section is research; a 10-response Google Form is homework. The methodology sections in each card are not decoration — they are the project.
The Arts & Science idea bank
Ratings are editorial judgments by FirstGrade (July 2026). Field studies involving people need faculty-approved consent protocols — build that into week one.
Starter kit — the measurement harness (works for every branch)
Almost every project on this site fails viva for the same reason: no honest evaluation. This kit gives you trial logging and the statistics your report needs — mean, standard deviation and a 95% confidence interval per condition — so "we tested it" becomes a table you can defend. Copy these four files into your repo and start logging from week one.
This harness records and analyses YOUR results — it cannot invent them. Do not fabricate rows: an examiner who asks you to re-run one condition in front of them will find out in about two minutes, and a fabricated results chapter is worse than an honest negative one.