Cohort Hiring as One Portfolio

A new research centre, a national priority programme or a strategic push into a field rarely needs one appointment. It needs six, or twelve, made over two or three years. They are often run as separate searches: each post advertised, each committee convened, each candidate judged on their own merits.

Sound individual decisions can still produce a poor cohort.

Clusters already form

Gulf universities already hire in clusters, whether or not they plan to. Of the 2,466 researchers who completed a move to a Gulf institution between 2021 and 2024, 48% share their leading subfield with another arrival at the same institution within a year, and 22% share the same narrow research topic. At the universities ATLAS models paper by paper, 48% of arrivals publish with another arrival who came within two years.¹

Some of that is planned. Some of it follows from separate searches drawing on the same visible pool.

Where one-at-a-time scoring fails

Scored alone, the strongest candidates for a set of posts often resemble each other: the same subfield, the same career stage, the same networks. Appointed together, they cover the same ground twice and leave the brief's other subjects uncovered.

Their effects also do not add up. Each appointment changes the institution's average, and the next appointment is measured against an average that has already moved. A cohort's effect has to be computed on the portfolio as a whole.

Exhibit 1· Illustrative

Two cohorts of four drawn from the same pool

	Cohort A: four strongest individual scores	Cohort B: planned as one portfolio
Subfields in the brief covered	2 of 4	4 of 4
Career stages	4 senior	1 senior, 2 mid-career, 1 early
Candidates who co-author with one another	3	1
Change in citation impact, 80% range	+0.06 to +0.21	+0.05 to +0.18

Cohort A has a slightly higher projected effect, but covers only two of the four subfields in the brief.

One decision, not several

A cohort is planned as one decision: which posts, at which career stages, and in what order. A senior appointment early in the sequence can anchor the posts that follow; an early-career post placed last can be filled from the anchor's own networks.

Each candidate is then scored twice: against the institution, and against the cohort as drawn so far, so that a second candidate in a covered subfield scores lower than a first. The cohort's combined effect on the institution's citation impact and its share of top-10% papers is estimated on the whole portfolio.²

The result is a Cluster Hiring Blueprint: the posts, their order, the field to be searched for each, and the expected effect of the cohort as a whole.

Footnotes

1. ATLAS movement registry, complete moves to Gulf institutions 2021 to 2024 (2,466 arrivals; 2,344 with a leading subfield), read 19 October 2026; counts are a floor. Subfield and topic: the leading OpenAlex subfield and topic of the papers each researcher published at the new institution. Co-authorship: 1,499 arrivals at universities ATLAS models paper by paper; a joint paper at the new institution with an arrival whose move began within two years. ↩

2. Ranges are 80% intervals. Impact values are on the ATLAS scale (world average 1.0). ↩