For Statisticians ·
What you'll accomplish
Defending a modeling choice, a missing-data approach, or an unusual test design often means finding the paper that already made the same call and explaining why. That search, done manually through a database, eats about an hour a week and repeats the same keyword-guessing every time. Elicit searches academic literature by meaning rather than exact keyword match, and then pulls method, sample size, and outcome details into a table you can scan in minutes instead of an hour.
What you'll need
What you should see: A clean search interface with a prompt asking what you want to find out, and a sidebar or menu pointing to search, reports, and your library.
Elicit works best with a focused question, not a topic name.
What you should see: A results list of papers ranked by relevance to your question's meaning, not just matching words. Troubleshooting: If the first results feel off-target, narrow the question further. A comparison framed as "A versus B for situation C" tends to return more usable results than a single broad term.
What you should see: A table with one row per paper and one column per field you asked for, each cell showing the extracted value with a link back to where it came from in the source paper. Troubleshooting: If a cell looks wrong or vague, click through to the source. Extraction quality varies by how clearly the paper reports that field.
Do not treat the extracted table as the final word.
What you should see: A confirmed match between what the table says and what the paper says, with a citation you could defend if someone asked you to pull the original.
What you should see: A running collection of verified sources, organized well enough that a future search on a related topic starts from something instead of nothing.