Explore the evidence corpus

How to use this corpus

Every paper here was read in full and its quantitative evidence recorded against the review's three questions.

Finding papers

The search box matches titles, authors and venues.

Forcing Sensitivity Choosing more than one question narrows the list: a paper must address every question you select. Forcing and Sensitivity together gives the papers that do both, not either.

2000s 2010s Year chips widen: any period you select counts. The number on each chip is that period's share of the whole corpus, so it stays put as you filter.

Selecting a question reveals a second segment on its chip, cycling through any, sufficient and partial. That requirement belongs to the question it sits on, so you can ask for solid Forcing evidence and merely partial Feedback in the same query. It is bound to the question because it means something different for each: most Forcing blocks carry sufficient evidence, while most Feedback blocks are partial, so a single strength control would flatter one question and gut another.

Reading a row

The badge names the questions a paper was classified under. The dots report what the extraction actually found, in Forcing / Sensitivity / Feedback order — colour says which question, shape says what was found: evidence found, partial, not addressed.

The two can differ, and that is not an error. A paper classified for Sensitivity often also reports the heat flux it imposes, so its Forcing dot is filled even though the badge does not mention Forcing.

Click any row to read its extracted evidence.

? opens this panel · Esc closes it