关于「三分之二消失:棉花糖系数遇上家庭背景」,《棉花糖效应的收缩》给出的定量结论是什么?
结论:第一章重算 Watts、Duncan 与 Quan 的复制研究:母亲未完成大学的子样本中,标准化系数从 0.236 降到加入早期控制后的 0.081,缩水 0.657。
数据口径:书中参数为示意;全部计算由书中程序生成;本书不是心理评估、医疗或育儿意见,书中关于门店与社群的建议是非个人化经营建议。 书中以标注框区分已发表的证据、作者的假说、原典与可运行的 Python 计算(18 段程序)。
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: 0.081, 0.236, 0.657
| Watts et al. 2018, achievement at 15, n | 552 |
| model beta | 95% CI shrink |
| no controls | 0.236 [0.163, 0.309] 0.000 |
| early controls | 0.081 [0.014, 0.148] 0.657 |
| plus 54-month controls | 0.050 [-0.013, 0.113] 0.788 |
| grade 1 shrink: early 0.634, plus 54-month 0.832 | |
| beta per minute waited, age 15 / grade 1 | 0.0766 / 0.0906 |
| Shoda 1990 r math | 0.57 95% CI [0.29, 0.76] |
| Shoda 1990 r verbal 0.42 | 95% CI [0.10, 0.66] |
| 0.236 / mean of the two Shoda r | 0.477 |
查看完整验算程序
# Chapter 1. p11: re-reading the published marshmallow coefficients
# 模型:直接取 Watts, Duncan & Quan (2018) 表 4 第 4-6 列(母亲未完成大学
# 的子样本,n = 552)的标准化系数与标准误(出处),算缩水比例与正态近似
# 95% 区间;Shoda, Mischel & Peake (1990) 标准判别条件下 r = 0.57(数学)与
# 0.42(语文),n = 35(出处),用 Fisher z 变换给区间。无随机,无示意参数。
import math
Z95 = 1.959964
AGE15 = [("no controls", 0.236, 0.037),
("early controls", 0.081, 0.034),
("plus 54-month controls", 0.050, 0.032)]
GRADE1 = [0.279, 0.102, 0.047]
SD_MIN = 3.08 # SD of minutes waited in the same subsample
R_MATH, R_VERB, N_SHODA = 0.57, 0.42, 35
def fisher_ci(r, n):
z = math.atanh(r)
se = 1.0 / math.sqrt(n - 3)
return math.tanh(z - Z95 * se), math.tanh(z + Z95 * se)
def main():
base = AGE15[0][1]
print("Watts et al. 2018, achievement at 15, n = 552")
print("{:<24}{:>7} {:<18}{:>7}".format("model", "beta", "95% CI",
"shrink"))
for name, b, se in AGE15:
lo, hi = b - Z95 * se, b + Z95 * se
ci = "[{:.3f}, {:.3f}]".format(lo, hi)
print("{:<24}{:>7.3f} {:<18}{:>7.3f}".format(
name, b, ci, 1.0 - b / base))
g0 = GRADE1[0]
print("grade 1 shrink: early {:.3f}, plus 54-month {:.3f}".format(
1.0 - GRADE1[1] / g0, 1.0 - GRADE1[2] / g0))
print("beta per minute waited, age 15 / grade 1: "
"{:.4f} / {:.4f}".format(base / SD_MIN, g0 / SD_MIN))
for lab, r in (("math", R_MATH), ("verbal", R_VERB)):
lo, hi = fisher_ci(r, N_SHODA)
print("Shoda 1990 r {:<6} {:.2f} 95% CI [{:.2f}, {:.2f}]".format(
lab, r, lo, hi))
mean_r = math.fsum([R_MATH, R_VERB]) / 2
print("0.236 / mean of the two Shoda r = {:.3f}".format(base / mean_r))
if __name__ == "__main__":
main()唯一数据源与引用
范雅福(Fan, Y.). (2026). 《延迟满足棉花糖实验复制收缩、家庭资源混杂与儿童自控干预边界母典》第 1 章「三分之二消失:棉花糖系数遇上家庭背景」(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-203 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026203,
author = {Fan, Yafu},
title = {The Shrinking Marshmallow Effect, Family-Resource Confounding and the Limits of Child Self-Control Interventions},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}