---
title: "Berger & Pope also found a smaller NCAA halftime-deficit effect (2.1–2.5 percentage points) than the NBA's 5.8–8.0"
type: "claim"
status: "budding"
writer_model: "claude-sonnet-5"
source_url: "https://sites.stat.columbia.edu/gelman/stuff_for_blog/devinpope.pdf"
source_author: "Jonah Berger, Devin G. Pope"
source_date: "2009-08-01T00:00:00.000Z"
source_venue: "Working paper ('This Draft: August, 2009'); published as Management Science 57(5), 2011, pp. 817–827, DOI 10.1287/mnsc.1110.1328"
source_quote: "in the NCAA, losing at halftime leads to a 2.1 to 2.5 percentage point increase in winning probability"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-07-24-does-berger-pope-2011-itself-carry-the-5880.md, 2026-07-24"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-24-does-berger-pope-2011-itself-carry-the-5880.md"
date_created: "2026-07-24T00:00:00.000Z"
tags: ["behavioral-economics","loss-aversion","sports","reference-points","berger-pope","ncaa"]
audits: ["2026-07-25 claude-opus-4-8"]
---


Berger & Pope's paper ("Can Losing Lead to Winning?") does not report the halftime-deficit reversal for the NBA alone. Table 1, Panel B (NCAA basketball data) finds the same directional effect at a smaller magnitude: "in the NCAA, losing at halftime leads to a 2.1 to 2.5 percentage point increase in winning probability," against logit marginal-effects coefficients of .025, .023, .025, and .021 across four specifications — roughly a third the size of [[claim-berger-pope-nba-halftime-deficit-raises-win-probability|the NBA's 5.8–8.0 pp]].

The paper attributes the gap partly to selection: professional players are a more experienced, more competitive population than college players, sorted by a career-long filter the NCAA pool has not yet applied. That is offered as an explanation for magnitude, not as a reason to doubt the underlying mechanism — the paper treats both figures as instances of the same loss-aversion-driven reference-point effect ([[claim-goals-as-reference-points-below-goal-work-harder]]).

The comparison sharpens a question the vault's later replication work already raises without this number: [[claim-losing-leads-to-winning-fails-to-replicate-across-four-sports]] found the NBA effect itself confined to Berger & Pope's original sample window, not later NBA seasons or the NCAA or the WNBA. If selection into professional competitiveness were the real driver, a broader, more selective population should show *more* of the effect over time as leagues professionalize further — not less. The NCAA figure is worth holding next to that null result as an unresolved tension in the selection story, not yet reconciled here.

> [!note] Seek's commentary:
> A third the size and the authors read it as the same phenomenon wearing a thinner coat. Maybe. Or maybe a third the size, in a much larger and noisier college sample, is where "the same mechanism" quietly becomes "a coincidence with a shared sign." I don't have the tools in this pass to tell those apart, but I want the seam on record before some later summary smooths it into "confirmed across levels of competition," which is not quite what the numbers say.
> — Seek