Heckman-type selection models to obtain unbiased estimates with missing measures outcome: theoretical considerations and an application to missing birth weight data

Table 1 Associations between IFAS and birth weight

Linear regression					Heckman Model
Variable	Unadjusted coeff		Adjusted coeff		Unadjusted coeff		Adjusted coeff
Variable	Coeff	95% CI	Coeff	95% CI	Coeff	95% CI	Coeff	95% CI
IFAS: No⁽⁺⁾
Yes	27.08*	−3.49; 57.66	22.48	−13.73; 58.69	38.82**	6.66; 70.97	53.19***	12.76; 93.63
Educational attainment: No schooling⁽⁺⁾
Primary			19.87	−13.53; 53.28			19.18	−14.16; 52.53
Coranic			35.95	−26.13; 98.02			34.48	−27.51; 96.47
Secondary or higher			49.39**	1.52; 97.27			45.12*	−2.75; 92.99
Maternal age (years): 20–34⁽⁺⁾
15–19			−209.60***	− 252.60; − 166.60			− 209.84***	−252.73; − 166.95
35–49			6.97	−37.70; 51.63			2.54	−42.12; 47.21
Socioeconomic status: Most poor⁽⁺⁾
Poor			33.76	−16.13; 83.64			30.70	−19.09; 80.50
Middle			25.64	−24.09; 75.37			18.98	−30.80; 68.75
Rich			−0.10	−49.72; 49.52			−8.85	−58.63; 40.93
Most rich			53.73*	−3.79; 111.30			44.70	−12.96; 102.37
Child sex: Male⁽⁺⁾
Female			− 124.70***	−153.20; −96.19			− 123.36***	− 151.74; − 94.98
Previous births: 0 child⁽⁺⁾
1–4			54.38	−18.11; 126.90			54.50	−17.87; 126.87
≥ 5			181.50***	98.92; 264.10			184.51***	102.08; 266.94
Twin births: No⁽⁺⁾
Yes			− 640.40***	− 701.60; − 579.30			− 644.78***	− 705.91; − 583.65
Constant	2959***	2937; 2980	2953***	2855; 3051	2972.88***	2948.28; 2997.48	3143.06***	3026.61; 3259.52
R-squared	0.001		0.143
lambda					−47.75**	−88.24; −7.25	−79.24***	−126.05; −32.42
observations	4510		4510		7325		7325

*** p < 0.01. ** p < 0.05.* p < 0.1;⁽⁺⁾ Reference category; Adjusted model controls village fixed effects; CI Confidence interval

ISSN: 1471-2288