关于「门店实验的设计:不偷看、不早停、先配对、按模拟定天数」,《门店小实验的重抽样推断》给出的定量结论是什么?
结论:第六章讲不偷看、不早停与配对:每天看一次使误报率升到 0.1714,按周配对的精确符号翻转检验 p 为 0.0234,不配对则为 0.8008。
数据口径:书中参数为示意;全部计算由书中程序生成;本书属于非个人化经营建议。 书中以标注框区分已发表的证据、可运行的 Python 计算(20 段程序)、原典与作者的假说。
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: 0.0234, 0.8008
| weekly differences: 55 30 -12 70 25 48 5 38 | |
| mean difference 32.38 | |
| paired sign-flip exact: 6 of 256 | p = 0.0234 |
| unpaired exact: 10306 of 12870 | p = 0.8008 |
| paired t = 3.432 | p = 0.0109 |
| sd between weeks 251.0 | sd of differences 26.7 |
查看完整验算程序
# Chapter 6. p63: pair what varies, then flip signs exactly.
# Model: two near-identical counters in one shop for 8 weeks; each
# week a coin decides which counter shows the new display. Weekly
# takings (yuan, illustrative) swing with holidays, so weeks differ far
# more than counters do. Paired analysis: under H0 the sign of each
# weekly difference is a fair coin, so all 2^8 = 256 sign patterns are
# equally likely (exact sign-flip test). Unpaired analysis of the same
# 16 numbers enumerates all C(16, 8) = 12870 splits. Paired t (p11).
import itertools
import math
def betacf(a, b, x):
tiny = 1e-300
qab, qap, qam = a + b, a + 1.0, a - 1.0
c = 1.0
d = 1.0 - qab * x / qap
d = 1.0 / (d if abs(d) > tiny else tiny)
h = d
for m in range(1, 300):
m2 = 2 * m
aa = m * (b - m) * x / ((qam + m2) * (a + m2))
d = 1.0 + aa * d
d = 1.0 / (d if abs(d) > tiny else tiny)
c = 1.0 + aa / c
c = c if abs(c) > tiny else tiny
h *= d * c
aa = -(a + m) * (qab + m) * x / ((a + m2) * (qap + m2))
d = 1.0 + aa * d
d = 1.0 / (d if abs(d) > tiny else tiny)
c = 1.0 + aa / c
c = c if abs(c) > tiny else tiny
de = d * c
h *= de
if abs(de - 1.0) < 1e-15:
break
return h
def betai(a, b, x):
if x <= 0.0:
return 0.0
if x >= 1.0:
return 1.0
lb = (math.lgamma(a + b) - math.lgamma(a) - math.lgamma(b)
+ a * math.log(x) + b * math.log(1.0 - x))
bt = math.exp(lb)
if x < (a + 1.0) / (a + b + 2.0):
return bt * betacf(a, b, x) / a
return 1.0 - bt * betacf(b, a, 1.0 - x) / b
def t_two_tail(t, v):
return betai(v / 2.0, 0.5, v / (v + t * t))
def t_cdf(t, v):
half = 0.5 * t_two_tail(t, v)
return 1.0 - half if t > 0 else half
def t_crit(alpha, v):
lo, hi = 0.0, 1000.0
for _ in range(200):
mid = 0.5 * (lo + hi)
if t_two_tail(mid, v) > alpha:
lo = mid
else:
hi = mid
return 0.5 * (lo + hi)
OLD = [820, 1150, 900, 1480, 760, 1010, 1290, 880]
NEW = [875, 1180, 888, 1550, 785, 1058, 1295, 918]
if __name__ == "__main__":
d = [b - a for a, b in zip(OLD, NEW)]
n = len(d)
obs = math.fsum(d)
ge = 0
for signs in itertools.product((1, -1), repeat=n):
ge += abs(math.fsum(s * v for s, v in zip(signs, d))) >= abs(obs)
print("weekly differences: {}".format(" ".join(str(v) for v in d)))
print("mean difference {:.2f}".format(obs / n))
print("paired sign-flip exact: {} of {} p = {:.4f}".format(
ge, 2 ** n, ge / 2 ** n))
pool = OLD + NEW
tot = math.fsum(pool)
d0 = abs(math.fsum(NEW) - math.fsum(OLD))
ge2 = cnt = 0
for idx in itertools.combinations(range(16), 8):
s = math.fsum(pool[i] for i in idx)
cnt += 1
ge2 += abs(2 * s - tot) >= d0 - 1e-9
print("unpaired exact: {} of {} p = {:.4f}".format(ge2, cnt, ge2 / cnt))
m = obs / n
sd = math.sqrt(math.fsum((v - m) ** 2 for v in d) / (n - 1))
t = m / (sd / math.sqrt(n))
print("paired t = {:.3f} p = {:.4f}".format(t, t_two_tail(t, n - 1)))
sdw = math.sqrt(math.fsum((v - math.fsum(OLD) / n) ** 2 for v in OLD)
/ (n - 1))
print("sd between weeks {:.1f} sd of differences {:.1f}".format(sdw, sd))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《自助法与置换检验、小样本门店实验不靠正态假设的重抽样推断母典》第 6 章「门店实验的设计:不偷看、不早停、先配对、按模拟定天数」(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-308 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026308,
author = {Fan, Yafu},
title = {The Bootstrap and Permutation Tests: Resampling Inference for Small Shop Experiments without Normality Assumptions},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}