Methodology & Assumptions

Overview

This directory tracks tenure-track Operations & Supply Chain Management (OM/SCM) faculty and ranks schools, PhD programs, and individuals on their research productivity and impact. This page documents the assumptions behind those numbers. Exact threshold values are shown in the Key parameters box above and are read directly from the live configuration.

Who is part of the directory

The directory focuses on business school faculty. Industrial Engineering departments are currently not tracked. The starting point for building the directory was the list of UTD 100 business schools. I subtracted a few schools from the list if they did not have a clearly distinguishable OM/SCM department, and added many schools, particularly outside of the United States.

In terms of areas within a school, I focused on operations management, operations research, supply chain management and decision analysis. Information systems faculty are currently not tracked.

The directory focuses on active faculty members. Emeritus and retired faculty members are not tracked. Only tenure track faculty are included. Faculty that request privacy have their information hidden. Note that whether a faculty member is tenure track or not is determined by their title on their department web page, which can lead to classification errors.

Data sources

• Publications and citations come primarily from OpenAlex. Where a faculty member is better represented by Google Scholar, that source is used instead. Each record links to a specific author profile; we cache the full per-paper history (title, year, venue, citation count, co-authors).
• A data pull is a refresh of this publication data from the sources above. The date of the most recent pull is shown in the Key parameters box; all time-based metrics (e.g. citations per year) are measured relative to the pull year.
• Schools are seeded from the UTD Top-100 Business School Research Rankings set.
• The ground truth for faculty information are department faculty web pages.

Journals counted

• Productivity and impact are computed over the UTD-24 list of premier business journals.
• A broader OM journal list is surfaced on faculty pages for context but is not part of the ranking metrics. Practitioner outlets are excluded from the metrics.
• Journal names are matched case-insensitively after normalizing punctuation.

Productivity metric

For each faculty member, productivity is:

UTD-24 publications ÷ (data-pull year − PhD year) — i.e. premier-journal output per year since the PhD.

• A school's productivity is the median of this value across its tenure-track OM/SCM faculty; a PhD program's is the median across its alumni in the directory.
• The faculty productivity leaderboard requires a minimum number of years since the PhD (see Key parameters) so that very recent graduates are not ranked on a tiny denominator.
• Productivity does not take into account the number of authors on a paper; it also does not factor in leaves and extensions granted to faculty members.

Impact metric

Impact is built on citations per year (CPY) of a paper:

CPY = citations ÷ max(1, data-pull year − publication year) — current-year papers get a one-year exposure floor.

• A faculty member's impact is a fractional h-index over the CPY of their UTD-24 papers. The integer part is the classic h-index — the largest h such that h of their papers are each cited at least h times per year — and the fraction interpolates toward the next level (see below). It is only shown once they have at least the minimum number of UTD papers (Key parameters).
• A school's (or PhD program's) impact is a fractional faculty-level h-index: the largest k such that k of its tenure-track OM/SCM faculty (or alumni) each have a personal impact score of at least k, again interpolated. This rewards genuine depth of high-impact faculty while not being inflated by a large roster of low-impact ones, and it cannot be carried by a single superstar.

Both levels use the same fractional (interpolated) h-index, described next.

Why a fractional h-index

A plain integer h-index produces heavy ties: dozens of faculty all sit at h = 10 and share a single rank, and the same happens to schools. The fractional h-index turns that step function into a continuous value in the range [h, h+1), so entities are ranked individually and the displayed number itself reflects how close they are to reaching the next whole level.

How it is computed

Sort the values — a faculty member's paper CPYs, or a school's per-member impact scores — in descending order as v1 ≥ v2 ≥ … ≥ vn. Take the integer h-index h (the largest h with v_h ≥ h). Then interpolate between the two boundary values: the one that qualifies (rank h) and the one that just misses (rank h+1). Drawing the straight line between those two points and taking where it crosses the diagonal y = x gives:

h_frac = (v_h + d·h) / (1 + d), where d = v_h − v_(h+1) is the gap between the two boundary values.

The result always lies in [h, h+1): if the next value is far below the bar, h_frac stays near h; if it nearly qualifies, h_frac approaches h+1. When every value qualifies (there is no entry beyond rank h), h_frac = h.

Worked example. Suppose a faculty member's paper CPYs, sorted, are 40, 22, 15, 11, 9.5, 8, 4. The integer h-index is 6 (the 6th value, 8, is ≥ 6, but the 7th, 4, is < 7). With v_h = 8 and v_(h+1) = 4, the gap is d = 4, so h_frac = (8 + 4·6) / (1 + 4) = 32 / 5 =6.4. Had that 7th paper instead had a CPY of 6 (closer to qualifying), d = 2 and h_frac = (8 + 2·6) / 3 =6.67 — a higher score, reflecting that the next level is within closer reach.

Subfield rankings

The Subfield Rankings page ranks faculty within a research subfield — a topical tag (e.g. Healthcare, Supply chain) or a methodological tag (e.g. Behavioral, Empirical).

• Each paper's tags are assigned by an LLM from its title/abstract against a closed taxonomy, deliberately conservative (up to 3 topical and 1 methodological tag per paper). Papers without an abstract — disproportionately older ones — are never tagged, so subfield paper counts undercount early work; treat these boards as indicative rather than exhaustive.
• A faculty member's subfield impact is the same fractional CPY h-index as the main impact metric, computed only over their UTD-24 papers carrying the selected tag, and shown once they have at least the minimum number of tagged UTD papers (same floor as the impact metric).
• Eligibility mirrors the main impact leaderboard (tenure-track OM/SCM, minimum post-PhD years); the top 30 per tag are shown.

Career trajectory charts

Faculty pages show trajectory sparklines plotting the productivity and impact metrics as they would have been computed at the end of each calendar year since the PhD — the same formulas as the headline numbers, but evaluated on the publications and citations on record up to that year. Each curve's right-hand endpoint equals the value shown elsewhere on the page.

• Productivity by career year restricts the UTD-24 publication count to papers appearing by that year, showing how premier-journal output per post-PhD year built up (and, for some, plateaued or declined).
• Impact by career year recomputes the CPY h-index over the UTD-24 papers published by that year, rewinding each paper's citations to the year in question (anchored on its lifetime total, with later-year citations from OpenAlex's per-year history subtracted off). Because that per-year history is bounded, early-career years are approximate for older papers; the present-day endpoint is exact.

When a tenure year is on record, a dashed vertical line marks it on both charts, so the pre- and post-tenure phases can be read off directly. The charts appear only for faculty with a recorded PhD year (and, for impact, per-year citation data); the tenure line only when a tenure year is known.

Collaboration statistics

Faculty pages show a collaboration card summarizing the co-author network. All of its statistics are computed over the same paper basis as the productivity metric: UTD-24 and OM journal papers only (a broad OpenAlex profile can pick up falsely attributed papers, and their co-authors would pollute the network), excluding display-only entries and papers with more than 25 authors. The card appears only for faculty whose publication source is OpenAlex — Google Scholar author strings include the faculty member themselves and are often truncated, and Scopus records carry no co-author lists.

Co-author identities are resolved by OpenAlex author ID where available. When a co-author appears without an ID on some papers, the occurrence is merged into their ID if the normalized name maps to exactly one known ID across the faculty member's papers; otherwise it is counted under the normalized name. Split OpenAlex profiles (one person under two IDs) are counted twice.

• Distinct co-authors is the number of unique co-author identities across all eligible papers.
• Repeat collaborators is the share of those distinct co-authors with at least two joint papers.
• New co-authors on each paper measures team stability: walking the eligible papers in chronological order, it is the fraction of co-author slots filled by someone the faculty member had not published with before. The first paper only seeds the "already seen" set — counting it would push every single-paper author to a meaningless 100% — so the ratio is taken over papers two onward, and is omitted entirely for faculty with fewer than two eligible papers. A low value indicates a stable recurring team; a high value indicates frequently changing co-author sets.
• Median team size is the median number of authors per eligible paper, including the faculty member.

As a reference point for interpreting an individual card, the directory-wide averages across the 2,422 faculty currently showing a collaboration card are: 21 distinct co-authors, 25% repeat collaborators, 65% new co-authors on each paper, and a median team size of 3.1. These averages are recomputed on every build.

Collaboration map

The Collaboration Map places every school in the directory on a world map and draws the co-authorship network between them.

• Locations come from each school's ROR record (GeoNames-backed coordinates), fetched once and cached. The map uses the Natural Earth projection.
• Circles are one uniform size; circle color is the number of active tenure-track OM/SCM faculty at the school (grey → green). The school's impact median appears in the hover tooltip.
• INSEAD is shown once per campus. Faculty are assigned to Fontainebleau, Singapore, Abu Dhabi, or San Francisco using the campus filters on INSEAD's own faculty listing; when someone appears on several campus listings the map uses the first of Fontainebleau > Singapore > Abu Dhabi > San Francisco. Only campuses with at least one active tenure-track OM faculty are drawn, and all campus circles link to the single INSEAD page.
• Arcs connect pairs of schools whose faculty co-authored papers in the last five years. Only co-authorships between directory faculty are counted (external co-authors carry no affiliation data), on the same paper basis as the collaboration statistics above: UTD-24 and OM journal papers, excluding display-only entries and papers with more than 25 authors. Each paper counts once per school pair. An arc's weight sums each shared paper's citations per year (citations divided by years since publication), so a recent well-cited collaboration outweighs an old or uncited one.
• Region flows (wide translucent bands) aggregate all cross-region school pairs into region-to-region totals — US census regions, Canada, Europe, Middle East, East / Southeast / South Asia, Oceania — so the long-distance structure stays readable. By default, cross-region ties are shown only as these bands; clicking a band expands it into its individual school-to-school arcs (or the aggregation can be switched off entirely). Same-region arcs are governed by the strength slider, which initially shows the strongest 15%; arc opacity and width scale with weight.
• PhD placement network (the map's second mode): violet arrows point from the PhD-granting university to the school where its graduates now hold a tenure-track OM/SCM position, weighted by headcount. All currently active faculty count, with no time window. PhD institutions are resolved through the same alias mapping as the PhD-program leaderboard and joined to directory schools by exact name — programs outside the directory are not drawn. Placements into one's own PhD school are excluded. Cross-region moves aggregate into directed region bands (both directions of a corridor can exist); clicking a band expands only that direction.

Collaboration centrality and the Collaboration Hub badge

Each school's collaboration centrality measures how essential it is to research collaboration in the field: its eigenvector centrality in the school co-authorship network described above (co-authored UTD-24 / OM-journal papers from the last five years, edges weighted by the papers' citations per year, co-authorships between directory faculty only). A school therefore scores high by collaborating with many schools that are themselves well connected. Edge weights are log-dampened before the computation so breadth across many partners counts for more than a single very strong partnership. Scores are scaled so the most central school reads 100; schools with no directory co-authorships in the window score 0. Ranks use competition ranking (tied scores share a rank).

The top 10 schools by collaboration centrality earn the Collaboration Hub award badge (ties at the boundary can push the number of holders above 10). Every school's score and rank appear on its page; the full ranking is the Collaboration toggle on the school leaderboard, which — unlike the productivity and impact rankings — applies no minimum-faculty floor, and displays the top 100 (school-page ranks are global).

Leaderboard display & ranking

• School and PhD leaderboards offer a toggle between productivity and impact.
• Rankings use competition ranking: tied entities share a rank, and the next distinct value skips ahead. The number of rows displayed is capped (Key parameters); medals are awarded to the top 30 on each metric.

PhD-program attribution

• Faculty are grouped into PhD programs by their doctorate-granting institution, after canonicalizing name variants (e.g. school-of-management names) to a single institution.
• A program needs at least the minimum number of alumni in the directory (Key parameters) to appear on the PhD leaderboard.

Reviewer Finder — matching by abstract

The Reviewer Finder helps editors locate conflict-free potential reviewers for a manuscript. Alongside matching by topical/methodological tags, it offers a by abstract mode: paste a manuscript's title and abstract and it returns directory faculty ranked by how closely their own published work resembles it.

• How the matching works. Every UTD-24 / OM-journal paper with an available abstract is converted once into a sentence embedding — a list of numbers that captures the abstract's meaning — using a compact open language model (a MiniLM sentence transformer). When you submit a manuscript, the same model embeds your abstract and each faculty member is scored by the cosine similarity between your manuscript and their single closest paper. A score near 1 means the manuscript and that paper sit very close together in meaning; the ranked list is sorted by this score.
• Nothing leaves your browser. The language model runs entirely client-side: the pasted manuscript is embedded locally and compared against precomputed paper vectors that ship with the site. No abstract text is uploaded to any server.
• Conflicts of interest are excluded exactly as in tag matching — anyone sharing an institution, a co-authorship, or a PhD program with a listed author is removed from the results.
• Optional restrictions. You can additionally narrow the ranked list to reviewers who work on selected topical/methodological tags and/or who have published in a named journal. These act as filters on top of the similarity ranking.

Limitations. Similarity is computed only over papers in the premier/OM-journal lists that have an abstract on record, so reviewers whose relevant work sits outside those journals, or whose abstracts are missing, may be under-ranked. The score reflects topical and semantic closeness, not a paper's quality or a reviewer's availability, and the embedding model is a small general-purpose one — the ranking is a starting point for editorial judgment, not a substitute for it.

Known limitations

• OpenAlex sometimes misses publications. I try to catch them through co-authors, but if a paper is missing from your record, claim your profile on OpenAlex, and keep it up to date.
• I have missed many departments and faculty. If you feel left out, my apologies. Hit the contact button and let me know. I will quickly remedy the situation.

Acknowledgements

I thank my wife, Min, for always supporting me, and my daughter Thalia, for being an amazing teenager during our time in Lisbon (where this site was created). I also thank my colleagues at Nova University for their warm welcome and support. I am grateful to the people of Portugal for welcoming us to their wonderful country for a year.

I would also like to thank Evgeny Kagan, Kostas Stouras, Blair Flicker, Charles Corbett, Jordan Tong, Ioannis Stamatopoulos, Beril Toktay, Elena Katok, Tinglong Dai, Atalay Atasu and the countless others who provided feedback and helped with getting this site up and running.