关于「实战工具与检查器规格:推理轨迹甄选台账、评分器、预注册与证伪」,《推理轨迹识人与结果运气剥离》给出的定量结论是什么?

结论:第八章给出实战工具与检查器规格:台账字段、五道闸门、蒸馏小工具「推理轨迹过程追踪与结果运气剥离评分器」、按四件套分组的检查条款与预注册方案,起约束作用的效度检验需要 347 人。一切推理轨迹采集须事先告知并取得同意,不写隐蔽测谎、诱导、欺骗或套取商业秘密,不以任何受保护特征作判别,结果成败单列、永不进入推理评分。书中以标注框区分已发表的证据、可运行的 Python 计算(21 段程序)、原典与作者的假说。

数据口径:事后解释常常是一套通俗理论:合成实验里,事后自述与真实权重的相关只有 -0.013,与旁观者凭常识的猜测却高达 0.826;同步出声思考保留了次序与跳跃,要求边做边解释则会改变被测过程。 核对文献后本书做了蓝图修正:人能表演推理,旁人判别真假的正确率只比抛硬币高一点,因此不主张「不可伪造的思维指纹」,只主张伪造成本随追问深度上升,并写出上下界与推翻条件;甄选效度以修订后的估计为锚,结构化访谈居首也只有 0.42 上下,单一方法有限,须按规则多方法合成。

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: 347

T1 validity r 0.25: n124 r 0.15: n = 347
T2 increment Delta R^2 0.0347, f^2 0.0482, n167
T3 depth 613 per group depth 4: 188 per group
T4 ICC 0.40 vs 0.60, k3: n = 53 people
binding size (T1 at r 0.15)347 people with 6-month follow-up
查看完整验算程序
# Chapter 8. p82: sample sizes for the pre-registered tests.
# T1 validity: reasoning score against the six-month delayed record,
#    Fisher z, two-sided alpha 0.05, power 0.80, r = 0.25 and 0.15.
# T2 increment over structured conversation and work sample:
#    Delta R^2 from p63 at rho 0.1 (0.528 vs 0.494), Cohen f^2, normal
#    approximation n = (za + zb)^2 / f^2 + predictors + 1.
# T3 forgery cost: flag rate of consented role-play performers against
#    genuine reasoners at depth 6 (0.784 vs 0.241) and depth 4 (0.218
#    vs 0.111), two proportions, per group.
# T4 stability: ICC(3,1) over three scenes, H0 0.40 against 0.60,
#    Fisher z for the intraclass correlation, one-sided alpha 0.05.
import math
from statistics import NormalDist

Z = NormalDist().inv_cdf


def n_corr(r, alpha=0.05, power=0.80):
    return math.ceil(((Z(1 - alpha / 2) + Z(power)) / math.atanh(r)) ** 2
                     + 3)


def n_two(p1, p2, alpha=0.05, power=0.80):
    pb = (p1 + p2) / 2
    num = Z(1 - alpha / 2) * math.sqrt(2 * pb * (1 - pb)) + Z(power) * \
        math.sqrt(p1 * (1 - p1) + p2 * (1 - p2))
    return math.ceil((num / (p1 - p2)) ** 2)


def zicc(rho, k):
    return 0.5 * math.log((1 + (k - 1) * rho) / (1 - rho))


def n_icc(r0, r1, k, alpha=0.05, power=0.80):
    d = zicc(r1, k) - zicc(r0, k)
    return math.ceil(2 + k / (2 * (k - 1)) *
                     ((Z(1 - alpha) + Z(power)) / d) ** 2)


def main():
    print("T1 validity   r 0.25: n = {}   r 0.15: n = {}".format(
        n_corr(0.25), n_corr(0.15)))
    r2_full, r2_two = 0.528 ** 2, 0.494 ** 2
    f2 = (r2_full - r2_two) / (1 - r2_full)
    n = math.ceil((Z(0.975) + Z(0.80)) ** 2 / f2 + 3 + 1)
    print("T2 increment  Delta R^2 {:.4f}, f^2 {:.4f}, n = {}".format(
        r2_full - r2_two, f2, n))
    print("T3 depth 6: {} per group   depth 4: {} per group".format(
        n_two(0.784, 0.241), n_two(0.218, 0.111)))
    print("T4 ICC 0.40 vs 0.60, k = 3: n = {} people".format(
        n_icc(0.40, 0.60, 3)))
    total = n_corr(0.15)
    print("binding size (T1 at r 0.15): {} people with 6-month "
          "follow-up".format(total))


main()

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《思维脉络溯源识人、过程追踪博弈与剥离意图结果假象母典》第 8 章「实战工具与检查器规格:推理轨迹甄选台账、评分器、预注册与证伪」(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-385 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026385,
  author    = {Fan, Yafu},
  title     = {Reasoning-Trajectory Profiling, Process-Tracing Games and Separating Cognitive Structure from Stated Intentions and Outcome Luck},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论