关于「150 的置信区间:从新皮层回归到八个模型的包络」,《邓巴分层与私域半衰期》给出的定量结论是什么?
结论:第一章重算 2021 年的重估:八个模型的 95% 区间从 2.1 延伸到 520.0,偏好模型给出 69.2 人、区间 3.8 到 292.0,同时容得下 15、50 与 150。
数据口径:书中的层级下限、半衰期、权重与停止概率均为示意参数;全部计算由书中程序生成;本书是非个人化经营建议,合规规则以各地现行法规为准。 书中以标注框区分已发表的证据、可运行的 Python 计算(19 段程序)、原典与作者的假说。
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: 15, 150, 2.1, 292.0, 3.8, 50
| model n est lo hi hi/lo xSE | 5 15 50 150 |
| Bayes nc+rest | 71 69.2 3.8 292.0 76.8 3.03 y y y y |
| Bayes nc ratio | 71 108.6 4.6 520.0 113.0 3.34 y y y y |
| Bayes brainwt | 142 79.8 4.1 329.9 80.5 3.06 y y y y |
| Bayes brainvol | 126 79.1 4.0 335.4 83.8 3.10 y y y y |
| GLS nc+rest | 71 16.4 2.1 127.7 60.8 2.85 y y y - |
| GLS nc ratio | 71 42.0 5.2 336.3 64.7 2.90 - y y y |
| GLS brainwt | 142 23.6 3.2 175.6 54.9 2.78 y y y y |
| GLS brainvol | 126 23.8 3.2 177.8 55.6 2.79 y y y y |
| envelope of all 8 intervals | 2.1 to 520.0 |
| common to all 8 intervals | 5.2 to 127.7 |
| intervals containing | 5 : 7 of 8 |
| intervals containing | 15 : 8 of 8 |
| intervals containing | 50 : 8 of 8 |
| intervals containing 150 | 7 of 8 |
| Dunbar 148 / preferred 69.2 | 2.14 |
查看完整验算程序
# Chapter 1. p11: how wide the 95% intervals around "150" really are
# Data: the eight estimates of human group size and their 95% intervals
# in Table 1 of Lindenfors, Wartel and Lind (2021, Biology Letters):
# four Bayesian and four GLS phylogenetic regressions on primate data.
# For each model the program reports the fold width hi/lo, the implied
# multiplicative standard error on the log scale, and whether 5, 15, 50
# and 150 fall inside. It then intersects all eight intervals.
# Source of data: Lindenfors et al. 2021, Table 1; the rest is plain
# arithmetic with no illustrative parameters.
import math
ROWS = [
("Bayes nc+rest", 71, 69.2, 3.8, 292.0),
("Bayes nc ratio", 71, 108.6, 4.6, 520.0),
("Bayes brainwt", 142, 79.8, 4.1, 329.9),
("Bayes brainvol", 126, 79.1, 4.0, 335.4),
("GLS nc+rest", 71, 16.4, 2.1, 127.7),
("GLS nc ratio", 71, 42.0, 5.2, 336.3),
("GLS brainwt", 142, 23.6, 3.2, 175.6),
("GLS brainvol", 126, 23.8, 3.2, 177.8),
]
LAYERS = (5, 15, 50, 150)
def main():
print("model n est lo hi hi/lo xSE "
"5 15 50 150")
inside = {k: 0 for k in LAYERS}
for name, n, est, lo, hi in ROWS:
fold = hi / lo
se = (math.log(hi) - math.log(lo)) / (2 * 1.959964)
marks = []
for k in LAYERS:
ok = lo <= k <= hi
inside[k] = inside[k] + (1 if ok else 0)
marks.append("y" if ok else "-")
print("{:15s} {:4d} {:6.1f} {:5.1f} {:6.1f} {:5.1f} {:4.2f} "
"{} {} {} {}".format(name, n, est, lo, hi, fold,
math.exp(se), *marks))
los = [r[3] for r in ROWS]
his = [r[4] for r in ROWS]
print("envelope of all 8 intervals : {:.1f} to {:.1f}".format(
min(los), max(his)))
print("common to all 8 intervals : {:.1f} to {:.1f}".format(
max(los), min(his)))
for k in LAYERS:
print("intervals containing {:3d} : {} of 8".format(
k, inside[k]))
pref = ROWS[0]
print("Dunbar 148 / preferred 69.2 : {:.2f}".format(148 / pref[2]))
main()唯一数据源与引用
范雅福(Fan, Y.). (2026). 《邓巴数分层、关系维护时间预算与私域客户衰减半衰期母典》第 1 章「150 的置信区间:从新皮层回归到八个模型的包络」(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-231 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026231,
author = {Fan, Yafu},
title = {Dunbar's Layered Numbers, Relationship-Maintenance Time Budgets and the Decay Half-Life of Private-Channel Customers},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}