关于「实证闭环:统一倒数的 A/B、计算器与预注册方案」,《直播间集体下单的临界耦合》给出的定量结论是什么?

结论:第八章给出统一倒数的 A/B 闭环、小工具「直播间同步序参量计算器」与预注册方案(示意参数下每组 26 场,模拟检验力 0.907)。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。

数据口径:书中的观众、节奏、耦合、口令强度、库存与订单均为示意参数;全部计算由书中程序生成;本书是非个人化经营建议,只讨论真实订单,紧迫感只能来自真实库存与真实截止,节律部分只讲同步数学,不作任何身心功效断言。 第八章给出统一倒数的 A/B 闭环、小工具「直播间同步序参量计算器」与预注册方案(示意参数下每组 26 场,模拟检验力 0.907)。

English version

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: 0.907, 26

pilot: mean R2adj A 0.0728, B 0.0385
gap 0.0342, pooled sd 0.0438, z sum 2.8016
sessions per arm n26
simulated power at n26: 0.907
if the true gap were 0.02: n76 per arm
if the true gap were 0.03: n34 per arm
if the true gap were 0.05: n13 per arm
查看完整验算程序
# Chapter 8. p82: How many sessions the pre-registered A/B test needs.
# Model: the unit is a session; the outcome is its R2adj (p81). Arm A
# (unified countdown) has a cluster share drawn from N(0.30, 0.08), arm B
# from N(0.20, 0.08), clipped to [0, 1]; each session has 150 orders.
# A pilot of 400 simulated sessions per arm estimates the mean gap and
# the sd of R2adj; the normal-approximation size per arm is
#   n = 2 (z_0.975 + z_0.80)^2 sd^2 / gap^2.
# 300 simulated trials at that size (two-sided z test, 5 %) check power.
# Shares, sizes and spreads are illustrative; fixed seed.
import math
import random
import statistics

PERIOD, LEN, ORDERS = 300, 3600, 150
rng = random.Random(20261124)


def session(share):
    share = min(1.0, max(0.0, share))
    ph = []
    for _ in range(ORDERS):
        if rng.random() < share:
            ph.append(rng.vonmisesvariate(2 * math.pi * 45 / PERIOD, 4.0))
        else:
            ph.append(rng.uniform(0, 2 * math.pi))
    x = math.fsum(math.cos(a) for a in ph) / ORDERS
    y = math.fsum(math.sin(a) for a in ph) / ORDERS
    return (ORDERS * (x * x + y * y) - 1) / (ORDERS - 1)


def arm(mean, k):
    return [session(rng.gauss(mean, 0.08)) for _ in range(k)]


pa, pb = arm(0.30, 400), arm(0.20, 400)
gap = statistics.fmean(pa) - statistics.fmean(pb)
sd = math.sqrt((statistics.variance(pa) + statistics.variance(pb)) / 2)
z = statistics.NormalDist()
za, zb = z.inv_cdf(0.975), z.inv_cdf(0.80)
n = math.ceil(2 * (za + zb) ** 2 * sd * sd / (gap * gap))
print("pilot: mean R2adj A %.4f, B %.4f" % (statistics.fmean(pa),
                                           statistics.fmean(pb)))
print("gap %.4f, pooled sd %.4f, z sum %.4f" % (gap, sd, za + zb))
print("sessions per arm n = %d" % n)
hits = 0
for _ in range(300):
    a, b = arm(0.30, n), arm(0.20, n)
    se = math.sqrt((statistics.variance(a) + statistics.variance(b)) / n)
    if abs(statistics.fmean(a) - statistics.fmean(b)) / se > za:
        hits += 1
print("simulated power at n = %d: %.3f" % (n, hits / 300))
for g2 in (0.02, 0.03, 0.05):
    print("if the true gap were %.2f: n = %d per arm"
          % (g2, math.ceil(2 * (za + zb) ** 2 * sd * sd / (g2 * g2))))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《藏本耦合振子同步、百人直播间集体下单相变与声频节律夹带临界耦合母典》第 8 章「实证闭环:统一倒数的 A/B、计算器与预注册方案」(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-239 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026239,
  author    = {Fan, Yafu},
  title     = {Kuramoto Coupled-Oscillator Synchrony: the Collective-Ordering Phase Transition in Hundred-Viewer Livestreams and the Critical Coupling of Rhythmic Entrainment},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论