Table 2. Regression coefficients of reported BMI in three models.

	Raw dataset		Imputed dataset
	Boys N = 163	Girls N = 140	Boys N = 174	Girls N = 161
Correction by body-shape perception: model 1
Constant	2.17	1.21	1.63	1.46
α for(much) too thin	6.68	19.33	6.72	19.01
β for (much) too fat	−5.92	0.78	−3.13	0.31
γ for (much) too thin × reported BMI	0.57	−0.12	0.59	−0.12
δ for about the right weight × reported BMI	0.92	0.99	0.95	0.98
ζ for (much) too fat × reported BMI	1.21	0.99	1.11	1.00
R2	0.70	0.87
RMSE	1.70	1.35
Cross-validation RMSE	1.83	1.40
Correction by objective socioeconomic status: model 2
Constant	2.47	−2.66	2.54	−1.73
α for Low	−2.60	2.18	−2.78	0.89
β for High	−0.06	4.58	−1.17	3.09
γ for low × reported BMI	1.03	1.06	1.04	1.08
δ for middle × reported BMI	0.92	1.21	0.92	1.16
ζ for high × reported BMI	0.91	0.97	0.96	1.00
R2	0.67	0.85
RMSE	1.80	1.47
Cross-validation RMSE	1.88	1.51

Reported BMI is based on self-reported weight and height;

Source: Escapad Paris 2010, analytic subsample with non-respondents (n = 341).