关于「东西方原典考辨:《商君书·境内》的计数定爵与 1960 年的奖学金断点」,《会员门槛的断点回归》给出的定量结论是什么?
结论:规则甲之下,计数在 24 与 25 上的单位分别是 666 个与 675 个,两者之比 1.014,分布平滑。规则乙之下,25 上的单位跳到 2656 个,而 24 上只剩 65 个,比值 40.862;从 20 到 24 的各档都被掏空了一块,共有 1981 个单位被推到门槛上。
数据口径:落笔前本书核对原文并作蓝图修正:1960 年的论文研究的是奖学金荣誉证书而非会员体系,2008 年的指南是方法论文,二者都没有给出任何会员权益的效应量,书中的高估倍数全部来自固定种子的示意数据;命题也必须带边界陈述:只识别门槛处的局部效应,客人能精确凑单时设计失效,估计对带宽敏感,不宜用全局高阶多项式。 书中参数为示意;全部计算由书中程序生成。
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: 1.014, 1981, 20, 24, 25, 2656
| count | rule A (linear) | rule B (quota at 25) |
| 19 | 675 | 675 |
| 20 | 706 | 501 |
| 21 | 677 | 386 |
| 22 | 610 | 218 |
| 23 | 667 | 175 |
| 24 | 666 | 65 |
| 25 | 675 | 2656 |
| 26 | 644 | 644 |
| 27 | 722 | 722 |
| 28 | 632 | 632 |
| mass at Q / mass at Q - 1: A 1.014, B 40.862 | ||
| units pushed to the quota: 1981 of 20000 |
查看完整验算程序
# Chapter 7. p71: a quota on a counted score invites piling at the quota.
# Abstract model of a rank-for-count rule in the spirit of the Book of
# Lord Shang (illustrative numbers, no historical data): each of 20000
# units has a baseline count S ~ Poisson(m), m ~ Uniform(10, 40). Rule
# A pays a fixed reward per counted unit (a linear schedule). Rule B
# adds a quota Q = 25 for the unit leader's promotion: a unit whose
# baseline lies k = 1 to 5 short of Q makes up the gap with probability
# 0.9 - 0.15 (k - 1). We print the counts around Q under both rules, the
# ratio of mass at Q to mass at Q - 1, and how many units were pushed.
# Seed 20261171 for both rules (same baselines).
import math
import random
Q, N = 25, 20000
def poisson(rng, m):
# Knuth's product method, adequate for m <= 40
lim, k, p = math.exp(-m), 0, 1.0
while True:
p = p * rng.random()
if p <= lim:
return k
k = k + 1
def counts(quota):
rng = random.Random(20261171)
out, pushed = [], 0
for _ in range(N):
s = poisson(rng, rng.uniform(10.0, 40.0))
r = rng.random()
k = Q - s
if quota and 1 <= k <= 5 and r < 0.9 - 0.15 * (k - 1):
s = Q
pushed = pushed + 1
out.append(s)
return out, pushed
if __name__ == "__main__":
a, _ = counts(False)
b, pushed = counts(True)
print("count rule A (linear) rule B (quota at {})".format(Q))
for s in range(Q - 6, Q + 4):
print("{:5d} {:15d} {:19d}".format(
s, a.count(s), b.count(s)))
print("mass at Q / mass at Q - 1: A {:.3f}, B {:.3f}".format(
a.count(Q) / a.count(Q - 1), b.count(Q) / b.count(Q - 1)))
print("units pushed to the quota: {} of {}".format(pushed, N))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《断点回归设计与会员等级门槛、消费满额升级线两侧的局部因果识别母典》第 7 章「东西方原典考辨:《商君书·境内》的计数定爵与 1960 年的奖学金断点」(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-303 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026303,
author = {Fan, Yafu},
title = {Regression-Discontinuity Designs at Membership-Tier Thresholds: Local Causal Identification on Either Side of a Spend-Based Upgrade Line},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}