Multiple ordinal regression assessing how parameters hypothesised to reflect consensus predict a rank order of scientific domains (main test: I-physical, II-biological-hard, III-biological-soft, IV-social sciences; extended test: same as main, with mathematics and humanities at the two ends). For details on each parameter, see introduction and methods. Use of first person pronouns was measured through an interaction term with a dummy variable separating single and multi-authored papers – the main effects for these latter two are retained to ensure a hierarchically well-formulated model. [Data sourced from Thomson Reuters Web of Knowledge].