关于「两块五与一毛钱:价格标签的期望增益及其赢家诅咒」,《开放标签与价格标签》给出的定量结论是什么?
结论:第二章复算价格标签研究:被告知常规价者报告疼痛减轻的比例为 0.854,折扣价者为 0.610;精确枚举显示,每组 41 人的研究在真实差距较小时,显著结果平均会夸大 2.58 倍。
数据口径:书中参数为示意;全部计算由书中程序生成;本书不是医疗建议,不宣称任何产品或做法对身体或健康有任何作用,书中流程与工具为非个人化经营建议。 书中以标注框区分已发表证据、可运行的 Python 计算(21 段程序)、原典与作者假说。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 0.610, 0.854
| relief 2.50 | 0.854 (0.745 to 0.962) 0.10: 0.610 (0.460 to 0.759) |
| diff +0.244 (CI +0.060 to +0.428), chi2 p 0.0127, Fisher p 0.0236 | |
| worse | 2.50: 0.195 (0.074 to 0.316) 0.10: 0.439 (0.287 to 0.591) |
| diff -0.244 (CI -0.438 to -0.050), chi2 p 0.0176, Fisher p 0.0317 | |
| odds ratio 3.73 (CI 1.28 to 10.88) | |
| extra responders per 100 people told the higher price | 24.4 |
| price ratio 2.50 / 0.10 | 25, log ratio 3.219 |
查看完整验算程序
# Chapter 2. p21: The price-tag placebo, recomputed from 82 volunteers.
# Published inputs: Waber, Shiv, Carmon and Ariely (2008), JAMA research
# letter: identical placebo pills, regular price 2.50 dollars vs
# discounted 0.10 dollars, 41 per arm; pain reduction reported by
# 85.4 % (95% CI 74.6 to 96.2) vs 61.0 % (46.1 to 75.9), P 0.02;
# pain increase 19.5 % vs 43.9 %. Counts 35/41, 25/41, 8/41 and 18/41
# are our reconstruction; they reproduce every printed percentage.
# 价格标签安慰剂:复算比例、区间、风险差、比值比与精确检验。
import math
from statistics import NormalDist
Z = NormalDist()
Z975 = Z.inv_cdf(0.975)
N = 41
RELIEF = (35, 25)
WORSE = (8, 18)
def wald(k, n):
p = k / n
h = Z975 * math.sqrt(p * (1 - p) / n)
return p, p - h, p + h
def logc(n, k):
return math.lgamma(n + 1) - math.lgamma(k + 1) - math.lgamma(n - k + 1)
def fisher(a, c, n):
tot = a + c
lo, hi = max(0, tot - n), min(n, tot)
den = logc(2 * n, tot)
p = [math.exp(logc(n, k) + logc(n, tot - k) - den)
for k in range(lo, hi + 1)]
obs = p[a - lo]
return math.fsum(q for q in p if q <= obs * (1 + 1e-9))
def chi2_p(a, c, n):
pp = (a + c) / (2 * n)
z = (a / n - c / n) / math.sqrt(pp * (1 - pp) * 2 / n)
return 2 * (1 - Z.cdf(abs(z)))
for label, (a, c) in (("relief", RELIEF), ("worse", WORSE)):
pa, la, ha = wald(a, N)
pc, lc, hc = wald(c, N)
print("{:<6} 2.50: {:.3f} ({:.3f} to {:.3f}) 0.10: {:.3f}"
" ({:.3f} to {:.3f})".format(label, pa, la, ha, pc, lc, hc))
rd = pa - pc
se = math.sqrt(pa * (1 - pa) / N + pc * (1 - pc) / N)
print(" diff {:+.3f} (CI {:+.3f} to {:+.3f}), chi2 p {:.4f},"
" Fisher p {:.4f}".format(rd, rd - Z975 * se, rd + Z975 * se,
chi2_p(a, c, N), fisher(a, c, N)))
a, c = RELIEF
lor = math.log(a * (N - c) / (c * (N - a)))
slor = math.sqrt(1 / a + 1 / (N - a) + 1 / c + 1 / (N - c))
print("odds ratio {:.2f} (CI {:.2f} to {:.2f})".format(
math.exp(lor), math.exp(lor - Z975 * slor), math.exp(lor + Z975 * slor)))
print("extra responders per 100 people told the higher price: {:.1f}"
.format(100 * (a - c) / N))
print("price ratio 2.50 / 0.10 = {:.0f}, log ratio {:.3f}"
.format(2.50 / 0.10, math.log(2.50 / 0.10)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《开放标签安慰剂预测编码、价格标签期望增益与杯弓蛇影反安慰剂母典》第 2 章「两块五与一毛钱:价格标签的期望增益及其赢家诅咒」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-201 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026201,
author = {Fan, Yafu},
title = {Open-Label Placebos and Predictive Coding, Price-Tag Expectancy Gains and the Nocebo of the Bow in the Cup},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}