Randomization does not justify logistic regression

被引：95

作者：

Freedman, David A. ^{[1
]}

机构：

[1] Univ Calif Berkeley, Dept Stat, Berkeley, CA 94720 USA

来源：

STATISTICAL SCIENCE | 2008年 / 23卷 / 02期

关键词：

models; randomization; logistic regression; logit; average predicted probability;

D O I：

10.1214/08-STS262

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

The logit model is often used to analyze experimental data. However, randomization does not justify the model, so the usual estimators can be inconsistent. A consistent estimator is proposed. Neyman's non-parametric setup is used as a benchmark. In this setup, each subject has two potential if treated and the other if untreated; only one of the two re-responses, can be observed. Beside the mathematics, there are simulation results, a brief review of the literature, and some recommendations for practice.

引用

页码：237 / 249

页数：13

共 49 条

[1]

AMEMIYA T, 1981, J ECON LIT, V19, P1483

[2]

Amemiya Takeshi., 1985, Advanced Econometrics

[3]

[Anonymous], JASA

[4]

[Anonymous], INT STAT REV

[5]

[Anonymous], 1935, Supplement to the Journal of the Royal Statistical Society, DOI DOI 10.2307/2983637

[6]

[Anonymous], 1845, ACAD R SCI METR BR 2

[7]

ARIS EMD, 2000, P 5 INT C SOC SCI ME

[8]

Berk RA., 2004, Regression analysis: a constructive critique

[9] Application of the logistic function to bio-assay [J].

Berkson, J .

JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1944, 39 (227) :357-365

[10]

Brady HenryE., 2010, Rethinking Social Inquiry: Diverse Tools, Shared Standards, V2nd

← 1 2 3 4 5 →