This paper investigates whether athletic press conference media environments exhibit systematic negative cognitive priming (the Golem Effect) and whether such framing is explained by on-field team performance or operates as an entrenched cultural constant. Using a computational lexical analysis of 12,365 official NFL press conference questions across four major media markets (Cleveland, Philadelphia, San Francisco, and New England), we demonstrate that Cleveland media maintains an adversarial deficit framing rate of 13.00% (Confrontational-to-Promotional ratio: 1.32), representing a statistically significant deviation from other NFL baselines (z = 6.11, p = 1.02 x 10⁻⁹). Crucially, a natural experiment during the 2023 NFL regular season reveals that an 11-6 Coach of the Year playoff team in Cleveland faced 35% higher deficit scrutiny (12.84%) than a collapsing 4-13 last-place team in New England (9.47%, p = 0.018). Furthermore, Cleveland deficit framing proved completely invariant between an 11-6 playoff season and a 3-14 collapse (12.84% vs. 12.72%, p = 0.920). We conclude that chronic media deficit framing functions as a persistent environmental tax on franchise leadership and player retention.
I. Introduction & The 25+ Year Institutional Churn
Since returning to the National Football League in 1999, the Cleveland Browns have endured unprecedented leadership instability. Over a 25+ year span, the franchise has cycled through 13 head coaches including interims, 10 general managers or top football personnel chiefs, and started 42 distinct quarterbacks. While conventional sports commentary attributes this instability entirely to internal front-office mismanagement or draft failures, organizational sociology suggests that high-churn institutions cannot be understood without examining their surrounding informational and cognitive ecosystems.
Every regime arrives with genuine optimism; within eighteen months, press briefings shift heavily toward deficit framing and historical relitigation; by year three, the pressure cycle often accelerates organizational turnover. To understand why this cycle repeats across completely different coaching staffs, front offices, and ownership structures, we must evaluate the communicative climate surrounding daily preparation.
II. Theoretical Framework: The Golem Effect in Athletics
The psychological foundation of this study rests on the expectancy-performance paradigm established by Rosenthal and Jacobson (1968) and further formalized by Babad, Inbar, and Rosenthal (1982). While the Pygmalion Effect demonstrates that positive expectations enhance performance, the Golem Effect proves that low or adversarial expectations systematically suppress performance.
In professional athletics, press conferences are not merely passive reporting opportunities; they are primary environmental inputs. Athletes and coaches spend hours each week answering questions that frame their daily reality. When that reality is repeatedly framed through fragility, suspicion of hidden injuries, and inevitable collapse, it taxes executive functioning and erodes organizational trust.
Our central empirical hypothesis was straightforward: Cleveland sports media maintains a measurable, persistent Golem Effect that prevents the organization from achieving sustained success.
III. Corpus Construction & Provenance
To eliminate selection bias, transcripts were harvested directly from official team media portals (1RMG network) utilizing court-stenographer transcripts. The final analysis corpus comprises 761 transcripts and 12,365 verified questions across four markets:
- Cleveland Browns: 302 transcripts, 5,555 questions across the 2021, 2023, 2024, 2025, and 2026 seasons.
- Philadelphia Eagles: 194 transcripts, 2,527 questions across the Nick Sirianni era and visiting opponents.
- San Francisco 49ers: 173 transcripts, 2,720 questions across the Kyle Shanahan era.
- New England Patriots: 173 transcripts, 2,539 questions across the Bill Belichick and Jerod Mayo eras.
IV. Computational Lexical Methodology
Each question was scored using a deterministic scoring engine that analyzes lexical density across two opposing poles:
Deficit Framing: Lexical tokens indicating failure, regression, doubt, catastrophic injury consequences, division, or impending doom (e.g., "concern," "struggle," "doubt," "fall apart," "frustration," "collapse").
Promotional Framing: Lexical tokens indicating execution, growth, schematic adjustments, leadership stability, and positive trajectory (e.g., "opportunity," "progress," "development," "complement," "confidence").
V. Robustness & Counter-Argument Testing
We subjected the data to two critical robustness tests designed to evaluate common defenses raised by sports journalists:
Counter-Argument 1: "Cleveland reporters are tougher because the team loses more."
Empirical Refutation: The 2023 natural experiment tested this directly. When Bill Belichick suffered his worst season in 30 years (4-13), New England reporters exhibited a 9.47% deficit framing rate. When Kevin Stefanski won 11 games and NFL Coach of the Year, Cleveland reporters subjected him to a 12.84% deficit framing rate (Odds Ratio 1.41, p = 0.018). The media was 35% more adversarial toward an 11-win playoff coach than toward a four-win collapse.
Counter-Argument 2: "Questions reflect the injury toll, not cynical bias."
Empirical Refutation: We conducted a rigorous sensitivity analysis by completely removing all procedural injury terms (injury, hurt, pain, surgery, rehab, setback, miss time) from the scoring engine. Even with injury terms excluded, Cleveland's non-injury confrontational rate remains 7.64%. This represents a 61% higher rate than New England (4.75%, z = 3.89, p = 9.82 × 10⁻⁵) and 25% higher than San Francisco (6.11%, z = 2.04, p = 0.041). Furthermore, between Cleveland's 2023 season (starting 5 quarterbacks with multiple tackles on IR) and 2025 season, the total deficit framing rate was virtually identical (12.84% vs. 12.72%, z = 0.10, p = 0.920), proving that deficit framing does not mechanically track injury frequency.
VI. Reproducibility & Open Science
The full replication code is publicly available in the project repository via reproduce_all_metrics.py and scoring_engine.py. Every calculation reported in this study can be independently verified from the raw court-stenographer JSON files.