Researcher Personality
Each researcher in the topical-homophily cohort gets a four-letter type assembled from four career-stage-neutral axes. Cell counts give the population inside each of the 16 types; click a cell to filter the table to its members. The classification covers the … researchers with a homophily score (≥3 coauthors with semantic communities).
16 personality cells (4×4 grid)
Rows: Mover/Stayer × Heterophilous/Homophilous.
Columns: Broker/Local × Team/Solo.
Click a cell to filter the table; click again to clear.
Who collaborates with whom (16×16)
For each ordered pair of personality types, the heatmap shows the
coauthor edges between them. Switch between row share
(what fraction of type-i's edges land in type-j) and lift
(row share divided by type-j's population fraction; ×1 = chance).
·
Hover a cell for details.
Per-author profile
Default sort is by citations. Click a column header to re-rank.
The four binary axes are shown as separate columns; the combined
type is in the first column.
The Mover signal uses the §9 trajectory class when available
(≥3 five-year bins after truncating the partial 2025–29 window),
falls back to a simple Stayer/Mover cut on raw bins for authors
with ≥2 observable bins, and defaults to Stayer for
authors whose career history is too short to register movement.
The Bridge axis is binary on weighted betweenness centrality
(Broker iff bcw > 0); 78% of the cohort sits at bcw = 0 in this
32k-node graph, so brokerage is genuinely rare.