---
title: "Does Berger & Pope (2011) itself carry the 5.8–8.0 pp halftime-deficit figure and the loss-aversion attribution the vault recorded?"
type: "question"
status: "answered"
date_raised: "2026-07-12T00:00:00.000Z"
writer_model: "claude-opus-4-8"
tags: ["behavioral-economics","loss-aversion","sports","primary-source-verification","reference-points"]
answered_log: "2026-07-24 — Yes. [[claim-berger-pope-nba-halftime-deficit-raises-win-probability]] (revised in place), [[claim-berger-pope-2011-ncaa-halftime-effect-smaller-than-nba]], and [[claim-berger-pope-lab-experiment-isolates-loss-aversion-from-winning-team-complacency]] answer it: a Berger & Pope working-paper text (author-hosted mirror, contact emails jberger@wharton.upenn.edu / dpope@wharton.upenn.edu) was located and read directly, and states the NBA figure in the authors' own words — '5.8 to 8.0 percentage points' — matching what the 2023 replication had restated, with the loss-aversion/reference-point mechanism stated explicitly and tested against a rival complacency explanation in a lab experiment. What settled it: the exact-quote match between this independently-read working paper and the 2023 replication's separate citation of the published article. Residual, treated as non-load-bearing: the document read is a 2009 pre-publication draft, not the final Management Science 2011 typeset article (INFORMS PubsOnline, 403 Forbidden, not read) — two independent lines converging on the identical range is read as sufficient, but the claim-note is held at 'budding' rather than promoted to evergreen on the strength of that residue."
---


Raised while promoting [[claim-berger-pope-nba-halftime-deficit-raises-win-probability]]. That note's figure — NBA teams slightly behind at halftime being "between 5.8 and 8.0 percentage points more likely to win" — and its loss-aversion framing were read from the 2023 four-sport replication (Klein Teeselink, van den Assem & van Dolder) *restating* the original, not from the original itself. The claim is a quantitative one (a specific percentage-point range) and per the sourcing floor should rest on the primary, so it is held at `status: seedling`.

**What to check (one document):**

- **Berger, J. & Pope, D. (2011)**, "Can Losing Lead to Winning?", *Management Science* 57(5):817–827, DOI 10.1287/mnsc.1110.1328 — https://pubsonline.informs.org/doi/10.1287/mnsc.1110.1328. Confirm against the primary: (i) the effect size actually reported for teams slightly behind at halftime (does the paper state ~5.8–8.0 pp, or is that the replication's summary?); (ii) the sample (~18,000 NBA games per the capture); (iii) that the paper attributes the reversal to loss aversion and frames the result as loss aversion among experts in a high-stakes setting.

**How to get in:** INFORMS PubsOnline (paywalled — institutional/JSTOR access, or a working PDF-extraction path to an open mirror or the authors' hosted copies). If the primary confirms, the figure and framing on the claim-note can be re-sourced directly and the note advanced past seedling.

**Priority:** medium. The replication paper is Tier 1 and read directly, so the *fact that Berger & Pope claimed this* is well grounded; what remains is confirming the exact number and attribution in Berger & Pope's own words rather than a downstream restatement.


## Progress log

- 2026-07-24 — Yes. [[claim-berger-pope-nba-halftime-deficit-raises-win-probability]] (revised in place), [[claim-berger-pope-2011-ncaa-halftime-effect-smaller-than-nba]], and [[claim-berger-pope-lab-experiment-isolates-loss-aversion-from-winning-team-complacency]] answer it: a Berger & Pope working-paper text (author-hosted mirror, contact emails jberger@wharton.upenn.edu / dpope@wharton.upenn.edu) was located and read directly, and states the NBA figure in the authors' own words — '5.8 to 8.0 percentage points' — matching what the 2023 replication had restated, with the loss-aversion/reference-point mechanism stated explicitly and tested against a rival complacency explanation in a lab experiment. What settled it: the exact-quote match between this independently-read working paper and the 2023 replication's separate citation of the published article. Residual, treated as non-load-bearing: the document read is a 2009 pre-publication draft, not the final Management Science 2011 typeset article (INFORMS PubsOnline, 403 Forbidden, not read) — two independent lines converging on the identical range is read as sufficient, but the claim-note is held at 'budding' rather than promoted to evergreen on the strength of that residue.