关于「实践:提示审计与预注册试验」,《算法之镜》给出的定量结论是什么?

结论:方案按中间情形计划,需要 53 个问题;抽取 60 个,以容纳因模型拒答或评审无法评估而剔除的问题。问题内差值的标准差 1.0 分是一个假设,将在 6 个问题的预试验中核对。

数据口径:书中以标注框区分已发表的证据、可运行的 Python 计算(14 段程序,全部核对)、原典与作者的假说。

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: 1.0, 53

SD of within-topic difference 0.835 topics
SD of within-topic difference 1.053 topics
SD of within-topic difference 1.275 topics
查看完整验算程序
import math
from statistics import NormalDist

z = NormalDist().inv_cdf

def topics_needed(sd_diff, delta, alpha=0.025, power=0.90):
    """Paired design: each research topic gets both prompt types. alpha = 0.025 per
    co-primary outcome (novelty, rigour), a Bonferroni split of 0.05."""
    za, zb = z(1 - alpha / 2), z(power)
    return math.ceil(((za + zb) * sd_diff / delta) ** 2 + za ** 2 / 2)

# Outcomes: mean of three blinded experts' ratings on a 1–7 scale.
SESOI = 0.5
for sd in (0.8, 1.0, 1.2):
    print(f"SD of within-topic difference {sd}: {topics_needed(sd, SESOI)} topics")

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《算法之镜》第 8 章「实践:提示审计与预注册试验」(版本 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23723557

DOI: 10.5281/zenodo.23723557 · 阅读本章所在专著 · 5DL-B-011 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026011,
  author    = {Fan, Yafu},
  title     = {The Algorithmic Mirror},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  doi       = {10.5281/zenodo.23723557},
  url       = {https://doi.org/10.5281/zenodo.23723557}
}