关于「重叠的分布:效应量、重叠系数与按性别分轨的错配率」,《觉察路径分岔与画像分轨》给出的定量结论是什么?

结论:第一章说明小效应意味着大重叠:小效应上沿两组重叠系数仍有 0.861;示意参数下按性别分轨错配 0.426,按信度 0.80 的画像只有 0.147。

数据口径:书中参数为示意或合成;全部计算由书中程序生成;本书不构成医疗建议,不替代合格专业人员,不对开悟或任何境界作保证。 第一章说明小效应意味着大重叠:小效应上沿两组重叠系数仍有 0.861;示意参数下按性别分轨错配 0.426,按信度 0.80 的画像只有 0.147。

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: 0.147, 0.426, 0.80

A. learners simulated40,000 (seed 41701)
d on the A - E axis0.373, OVL = 0.852
B. rule mismatch mean fit loss (SD units)
by sex label0.426 0.422
self-choice r.50 0.246 0.157
profile r.80 0.147 0.057
C. closed form
by sex: Phi(-shift / sd)0.426
profile reliability 0.50: acos(sqrt rel) / pi0.250
profile reliability 0.70: acos(sqrt rel) / pi0.185
profile reliability 0.80: acos(sqrt rel) / pi0.148
profile reliability 0.90: acos(sqrt rel) / pi0.102
查看完整验算程序
# Chapter 1. p12: mismatch rate of sex-based versus profile-based tracks.
# Each learner has two starting tendencies, A (analytic decoupling) and
# E (embodied relational attunement), standard normal with r = 0.10.
# Group means (illustrative, inside the small band of Hyde 2005):
#   group F: A = -0.10, E = +0.15     group M: A = +0.10, E = -0.15
# so d_A = 0.20 and d_E = 0.30. The better-fitting track is analytic
# when A > E. Rules compared: by sex label; by a measured profile with
# reliability 0.80; by self-choice read as a profile of reliability 0.50.
# Mismatch = assigned track differs from the better-fitting track.
import math
import random

R_AE = 0.10
MEANS = {"F": (-0.10, 0.15), "M": (0.10, -0.15)}


def phi(x):
    return 0.5 * math.erfc(-x / math.sqrt(2.0))


def draw(rng, sex):
    ma, me = MEANS[sex]
    z1, z2 = rng.gauss(0.0, 1.0), rng.gauss(0.0, 1.0)
    a = ma + z1
    e = me + R_AE * z1 + math.sqrt(1.0 - R_AE ** 2) * z2
    return a, e


def noisy(rng, diff, rel):
    sd_true = math.sqrt(2.0 * (1.0 - R_AE))
    sd_err = sd_true * math.sqrt((1.0 - rel) / rel)
    return diff + rng.gauss(0.0, sd_err)


if __name__ == "__main__":
    rng = random.Random(41701)
    n = 20000
    miss = {"sex": 0, "profile": 0, "self": 0}
    loss = {"sex": [], "profile": [], "self": []}
    for i in range(2 * n):
        sex = "F" if i % 2 == 0 else "M"
        a, e = draw(rng, sex)
        diff = a - e
        best = diff > 0
        picks = {"sex": sex == "M",
                 "profile": noisy(rng, diff, 0.80) > 0,
                 "self": noisy(rng, diff, 0.50) > 0}
        for k, pick in picks.items():
            if pick != best:
                miss[k] += 1
                loss[k].append(abs(diff))
            else:
                loss[k].append(0.0)
    sd = math.sqrt(2.0 * (1.0 - R_AE))
    shift = 0.25
    print("A. learners simulated: {:,} (seed 41701)".format(2 * n))
    print("  d on the A - E axis = {:.3f}, OVL = {:.3f}".format(
        2 * shift / sd, 2.0 * phi(-shift / sd)))
    print("B. rule              mismatch   mean fit loss (SD units)")
    for k, label in (("sex", "by sex label"), ("self", "self-choice r=.50"),
                     ("profile", "profile r=.80")):
        print("  {:<18} {:.3f}      {:.3f}".format(
            label, miss[k] / (2.0 * n), math.fsum(loss[k]) / (2.0 * n)))
    print("C. closed form")
    print("  by sex: Phi(-shift / sd) = {:.3f}".format(phi(-shift / sd)))
    for rel in (0.50, 0.70, 0.80, 0.90):
        print("  profile reliability {:.2f}: acos(sqrt rel) / pi = {:.3f}"
              .format(rel, math.acos(math.sqrt(rel)) / math.pi))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《两性深度觉察路径分岔:抽象逻辑解构与具身关系照见的认知神经母典》第 1 章「重叠的分布:效应量、重叠系数与按性别分轨的错配率」(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-417 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026417,
  author    = {Fan, Yafu},
  title     = {Divergent Pathways of Contemplative Insight: Top-Down Analytical Decoupling versus Embodied Relational Attunement},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论