Table 3.

Variance components, ICCs, attenuation bias in logistic regression coefficients, and associated potential attenuation of odds ratios in a hypothetical epidemiologic study.

Chemical class/ pesticide	σ2BW (95% CI)	σ2WI (95% CI)	ICCa	Percent attenuation biasb	ORobs if ORtrue = 2.0c
Carbamates
Carbaryl	1.1 (–0.04, 2.3)	1.3 (0.73, 2.0)	0.45	–55	1.4
Propoxur	2.0 (0.41, 3.6)	0.75 (0.33, 1.2)	0.73	–27	1.7
Organochlorines
Chlordane	3.2 (1.2, 5.1)	0.16 (0.10, 0.23)	0.95	–5	1.9
Methoxychlor	1.8 (–0.31, 3.8)	1.5 (0.36, 2.7)	0.54	–46	1.5
Organophosphates
Chlorpyrifos	0.40 (0.01, 0.79)	0.43 (0.25, 0.61)	0.48	–52	1.4
Diazinon	3.0 (0.47, 5.5)	0.99 (0.46, 1.5)	0.75	–25	1.7
Pyrethroids
Cyfluthrin	1.6 (0.28, 2.8)	0.84 (0.48, 1.2)	0.65	–35	1.6
Cypermethrin	3.6 (0.97, 6.2)	0.74 (0.42, 1.1)	0.83	–17	1.8
Permethrin	2.4 (0.78, 4.0)	0.37 (0.22, 0.53)	0.87	–13	1.8
Synergist
Piperonyl butoxide	6.2 (1.5, 11)	0.99 (0.56, 1.4)	0.86	–14	1.8
Herbicides
Dacthal	2.2 (0.42, 3.9)	0.96 (0.35, 1.6)	0.69	–31	1.6
Simazine	1.3 (–0.36, 2.9)	2.2 (1.2, 3.1)	0.37	–63	1.3
Trifluralin	1.5 (0.44, 2.7)	0.32 (0.10, 0.54)	0.83	–17	1.8
Abbreviations: Abbreviations: σ2BW, between-home variance; σ2W, within-home variance; CI, confidence interval; ICC, intraclass correlation coefficient. aσ2BW/(σ2BW + σ2WI). b[(β1,obs – β1,true)/β1,true] × 100%. cORtrueICC.