关于「实证闭环:钻饰报价器、双轨文案与预注册检验」,《钻石魔术尺寸与培育钻崩价》给出的定量结论是什么?

结论:第八章给出钻饰报价器、双轨文案规则与预注册检验方案,门店对照试验每组需要 356 位顾客。书中以标注框区分已发表的证据、作者的假说与原典,共含 20 段程序。

数据口径:书中的价格、比例与成本参数,除注明出处者外均为示意;全部计算由书中程序生成;本书不是个人化的法律、理财或经营建议,涉及的标注与广告规则以上架前人工复核的现行法条为准。 第三章用切磨者模型说明台阶怎样在重量分布上留下缺口与堆积,示意模型中 1.00 克拉一档有 0.650 的石头是被推上关口的;并比较左位数注意力与焦点从众两种解释。

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: 356

(a) customers per arm for 0.30 vs 0.40: 356
(b) stones per side for a 5 % step: 95
(c) monthly observations to detect -0.05 a year: 33
    span 2.8 years, slope se 0.0176
refutation thresholds, fixed before data:
  step at 1.00 below 0.02 with upper 95% bound below 0.05
  lab-grown log-price slope above -0.05 a year over 3 years
查看完整验算程序
# Chapter 8. p82: Sample sizes for the pre-registered tests.
# (a) Shop A/B test: share of customers who accept the 0.9x offer,
#     0.30 (control: the list leads with 1.00 ct) vs 0.40 (treatment).
# (b) Magic-size step: n stones per side to detect a 5 % step with
#     residual sd 0.12 (p23), alpha 0.05 two-sided, power 0.80.
# (c) Lab-grown price trend: how many monthly price observations (same
#     specification) give a slope standard error small enough to tell
#     a decline of 0.05 a year in log price from zero, residual sd
#     0.08 a month. All rates are illustrative design targets.
import math
from statistics import NormalDist

Z = NormalDist()
za, zb = Z.inv_cdf(0.975), Z.inv_cdf(0.80)
p0, p1 = 0.30, 0.40
pb = (p0 + p1) / 2.0
n_ab = ((za * math.sqrt(2 * pb * (1 - pb))
         + zb * math.sqrt(p0 * (1 - p0) + p1 * (1 - p1))) ** 2
        / (p1 - p0) ** 2)
print("(a) customers per arm for 0.30 vs 0.40: {}".format(math.ceil(n_ab)))
d, s = math.log(1.05), 0.12
print("(b) stones per side for a 5 % step: {}".format(
    math.ceil(2 * (za + zb) ** 2 * s * s / (d * d))))
target = 0.05
for m in range(6, 200):
    ts = [i / 12.0 for i in range(m)]
    tb = math.fsum(ts) / m
    sxx = math.fsum((t - tb) ** 2 for t in ts)
    se = 0.08 / math.sqrt(sxx)
    if target / se >= za + zb:
        print("(c) monthly observations to detect -0.05 a year: {}"
              .format(m))
        print("    span {:.1f} years, slope se {:.4f}".format(m / 12.0, se))
        break
print("refutation thresholds, fixed before data:")
print("  step at 1.00 below 0.02 with upper 95% bound below 0.05")
print("  lab-grown log-price slope above -0.05 a year over 3 years")

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《钻石克拉魔术尺寸价格跳跃、培育钻学习曲线崩价与供给卡特尔信号母典》第 8 章「实证闭环:钻饰报价器、双轨文案与预注册检验」(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-209 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026209,
  author    = {Fan, Yafu},
  title     = {Diamond Magic-Size Price Jumps, the Lab-Grown Learning-Curve Slide and Supply-Cartel Signalling},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论