Methods and data
How the numbers on this site were made
The research counts on unfreeze.us come from three collections I built: the Unfrozen Research Index, a database of management and organization journal articles; the Atlas of the Middle, an evidence map of 173 studies from the Index on middle managers, 105 of them about change (as of 30 October 2026); and the Evidence Library, graded findings on leading change. Outside figures quoted on other pages, such as survey statistics on the Veterans page, are cited where they appear. This page says how each collection was made, what the counts mean, and where they are weak.
Three collections, three denominators
- The Unfrozen Research Index: 34,953 research articles and reviews from 29 journals, 2007 to 2026 (2026 to date). It is used for counting, such as how often abstracts mention middle managers. Its records are not graded.
- The Atlas of the Middle: 173 studies on middle managers as of 30 October 2026, 105 of them about change, published 2007 to 2026 apart from one 1996 study, and drawn from the Index. Version 1 held 117 studies, nearly all first published from 2007 to 2018; Version 2 added 56 from 2019 to 2026. Each has a first-pass strength grade, and no study in it is currently rated high strength.
- The Evidence Library: 2,367 graded findings as of October 2026. A finding is one result from one study or review. 58 rest on tests strong enough to rate High. Most grades come from abstracts; 284 have been checked against the full paper. It includes studies from fields outside management, such as health services, so its studies are not limited to the Index journals. The homepage evidence and the Loop's evidence lines come from it.
A study can appear in more than one collection. The counts answer different questions, so they do not add up or contradict each other: 58 of 2,367 is a share of Library findings, and 173 is a count of Atlas studies.
The Unfrozen Research Index
Articles from 29 management and organization journals, 2007 to 2026, collected from open scholarly sources: OpenAlex, Semantic Scholar, Crossref and CORE, plus a research file from my doctoral program. Cleaning is logged. Counts after cleaning are within 2007 to 2026 (Index v4d: snapshot of 12 October 2026, corrected 1 November 2026).
Keyword counts use the 24,443 abstracts supplied by OpenAlex (24,037 of them in English, the set the comparisons use), which cover 28 of the 29 journals (21 core, 7 expansion). OD Practitioner comes only from my doctoral research file and has no OpenAlex abstracts. An abstract counts as mentioning executives or middle managers if its title or abstract uses one of a fixed list of terms. Across all 28 journals from 2007 to 2026 (2026 to date), 193 English-language abstracts mention middle managers, against 1,541 that mention executives.
Counts that compare executives and middle managers use English-language abstracts only. In October 2026 I found 405 abstracts from the Spanish-language edition of Harvard Business Review (2007 to 2011, plus one from 2016) among the OpenAlex abstracts, and 1 more in another language. The English search terms matched those almost only through the word "CEO", which Spanish uses too (90 of 91 matches; the other was the English name "Corporate Executive Board"), and never through a middle-manager term, so they inflated the executive side. They are left out of the comparison (index snapshot of 12 October 2026, cleaning step C7). Before this fix the headline read 786 to 66.
OpenAlex also dated 23 Strategic Management Journal records from 1994 and 1996 special issues as 2007 (22 papers and the special issue's list of reviewers). Their DOIs give the volume, and so the print year, so on 1 November 2026 they were moved to their print years, which puts them outside every window here, and the special issue's list of reviewers was reclassified as front matter (cleaning step C6). Before that fix the headline read 695 to 66.
Abstracts obtained under library or API terms are used for counting and are not republished.
The headline count
In the 21 core journals from 2007 to 2018, abstracts that mention executives outnumber abstracts that mention middle managers 693 to 65, about 11 to 1. This is a keyword count on English-language abstracts from OpenAlex. Counting abstracts from other sources for the same 21 journals and years also gives a ratio near 11 to 1.
Keyword searches miss papers, so a recall audit coded a stratified sample of 520 abstracts against a written guide (511 once the Spanish-language ones are left out). Corrected for what keywords miss, the core-journal ratio is about 10 to 1, with a plausible range of roughly 6 to 15. The audit coding is provisional until an independent human coder checks half of it.
Year-by-year and journal-by-journal counts, with CSV downloads, are in the research note The missing middle.
My earlier doctoral research file (12,031 abstracts from 22 journals, 2007 to 2018) found 853 executive and 68 middle-manager abstracts. That search also included subject keywords. The index count of 693 to 65 replaces it as the headline, because the index can be rebuilt by anyone and extended past 2018. Both counts, drawn from different abstract sources over largely the same articles, find executives named far more often than middle managers.
The Atlas of the Middle
- Version 1: 196 candidate papers from 2007 to 2018 were found with a narrow middle-manager search plus a broader definition within change studies. 117 were judged to be about middle managers.
- Version 2 (30 October 2026): papers first published from 2019 to 2026 were drawn from the same index journals (core and expansion) and screened from their abstracts with the same coding guide. Of 71 candidates, 59 were judged to be about middle managers. Of those, 56 were added after two independent fact-check passes and my approval, and 3 were held back: one appears to reuse another study's interviews, one reports no results in its abstract, and one is an essay. Of the 56 added, 10 were checked against the full text and the rest against their abstracts.
- Each study was coded for design, setting, the parts of the Unfrozen Loop it informs, whether it supports, complicates or challenges each part, and how strong its evidence is. The first 117, in Version 1, were first coded as follows: 102 from the abstract alone, 10 from the full text and 5 from the abstract plus a labeled summary.
- For Version 1, a blind second coding pass, made with the same method as the first, recoded 30 papers: 97% agreement on relevance, and agreement by Loop part (Cohen's kappa) of 0.51, 0.85, 0.79, 0.76 and 1.00. This checks the method's consistency; it is not the independent human review described below. An audit pass then reviewed every link.
- For Version 1, in October 2026 every plain-language summary was checked twice against the study's abstract or full text, in two independent passes. Missing summaries were written, overstated ones were corrected, and citation years now follow the print issue. The same check flagged 18 Loop links whose fit is weak; they go to the independent human check.
What isn't settled
- No person has yet checked the codes independently. When an independent human check is done, any change it makes to the numbers will be published here.
- Agreement on the first Loop part (0.51) is only moderate.
- Most studies were coded from abstracts, which can leave out details that matter.
- OpenAlex publication years can be wrong. 23 Strategic Management Journal records from 1994 and 1996 special issues (22 papers and the special issue's list of reviewers) were dated 2007; they now carry their print years (cleaning step C6, 1 November 2026), which took 2 executive and 1 middle-manager mention out of the headline count. One of them, Noda and Bower (1996), is in the Atlas under its print year.
- Papers from 2019 to 2026 were added in Version 2, screened from abstracts with the same coding guide. The blind second coding pass above covered Version 1 only, and like the rest of the map, the Version 2 coding is provisional until an independent human coder checks it.
- A study can inform more than one Loop part, so the part counts (67, 42, 37, 27 and 6, as of 30 October 2026) add up to more than 173. Only 6 studies speak to reviewing a change or holding it steady, and across the whole map no study is currently rated high strength.
- This is an evidence map, not a systematic review or meta-analysis.
How to cite
Cite the version you used. Version 2 is current. Version 1 figures (117 studies, 74 of them about change) stay correct for anything published before 30 October 2026.
Lopez, G. (2026). The Atlas of the Middle: An evidence map of 173 studies on middle managers (Version 2, 30 October 2026). Unfrozen. https://unfreeze.us/atlas/
Lopez, G. (2026). The Atlas of the Middle: An evidence map of 117 studies on middle managers (Version 1, October 2026). Unfrozen. https://unfreeze.us/atlas/
Corrections
Found a mistake? Email info@unfreeze.us. Corrections are made and noted.