Table 3.

 Association between own height and friend’s height difficulties with multicollinearity and confounding*. Figures are odds ratios with robust standard errors

Variable	Col 1	Col 2	Col 3	Col 4
Lagged own height	0.89† (0.01)	—	0.88† (0.01)	0.881† (0.01)
Friend’s height	0.21† (0.03)	0.18† (0.022)	0.01‡ (0.01)	0.009 (0.01)
Lagged friend’s height	−0.21† (0.03)	—	—	—
Wave	1.37† (0.08)	0.11 (0.21)	1.678† (0.09)	1.580† (0.11)
Fixed effects
Observations	4284	4284	4284	4284
R2	0.91	0.53	0.91	0.91

*Additional controls: sex, age, race, maternal education, family income, grade level in wave 1, indicator for missing family information. Column 1 recreates previous specifications by controlling both own and friend’s lagged height. Column 2 repeats column 1 except lagged own height and lagged friend’s height are not controlled. Column 3 repeats column 2 with the addition of lagged own height. Column 4 repeats column 3 with the addition of controls for environmental confounding in the form of school level fixed effects.

†P<0.001.

‡P<0.1.