Skip to main content

Table 2 Variables in the models with GLM (linear regression, lognormal regression, gamma regression, regression with Box Cox power transformation), GAMLSS, quantile regression for τ = 0.9 (QR 0.9) and for τ = 0.97 (QR 0.97) for the School Entry Health Examination Study data in Bavaria, 2001–2002.

From: Alternative regression models to assess increase in childhood BMI

  GLM GAMLSS QR 0.9 QR 0.97
Sex (SEX) + + + 0
Pregnancy smoking (PS) 0 [0] + 0
Watching TV (TV) + + + +
Breastfeeding (BF) 0 (+) + +
Meal frequency (MF) + + + +
Parents' graduate (PG) 0 [0] 0 0
Age (AGE) + + + 0
Maternal BMI (MB) + + + +
Weight gain in first 2 y (WG) + + + +
  1. "+" denoting significant variables, "0" non-significant variables and, in case of GAMLSS, "(+)" variables only significant for the σ term and " [0]" variables only significant in the univariate models.5