Skip to main content
. 2018 Dec 5;5(2):2055102918816606. doi: 10.1177/2055102918816606

Table 2.

Multilevel model estimates.

Predictors Model 2 (effects of time)
Model 3 (covariates)
Model 4 (final)
b (SE) t (df) b (SE) t (df) b (SE) t (df)
Intercept 0.05 (0.01) 15.27 (224)* 0.05 (0.01) 15.01 (220)* 0.05 (0.01) 15.33 (217)*
 Internet versus email group <0.01 (0.01) 0.75 (220) <0.01 (0.01) 0.63 (217)
 Internet versus discussion group <–0.01 (0.01) –0.66 (220) <–0.01 (0.01) –0.47 (217)
 BMI –0.01 (0.01) –1.56 (220) <–0.01 (0.01) –1.37 (217)
 Age <0.01 (0.01) 1.02 (220) <0.01 (0.01) 1.00 (217)
 Health reason <0.01 (0.01) 0.53 (217)
 Appearance to others <–0.01 (0.01) –0.06 (217)
 Appearance to self –0.01 (0.01) –1.48 (217)
–PRPE %: χ 2 (∆ df) 67.94% 567.02 (5) 0.45% 14.29 (4)* 0.46% 33.75 (3)*
Change over time (linear) –0.03 (0.01) –2.31 (224)* –0.03 (0.01) –2.60 (220)* –0.03 (0.01) –2.91 (217)*
 Internet versus email group 0.04 (0.02) 2.13 (220)* 0.03 (0.02) 1.90 (217)
 Internet versus discussion group –0.03 (0.02) –1.83 (220) –0.03 (0.02) –1.57 (217)
 BMI <0.01 (0.01) 0.06 (220) 0.01 (0.01) 0.67 (217)
 Age –0.03 (0.01) 2.29 (220)* 0.04 (0.01) 3.00 (217)*
 Health reason –0.02 (0.01) –1.60 (217)
 Appearance to others 0.02 (0.02) 1.07 (217)
 Appearance to self –0.04 (0.02) –2.31 (217)*
–PRPE %: χ 2 (∆ df) 3.20% 17.55 (4)* 3.49% 18.63 (3)*
Change over time (curvilinear) <0.01 (0.01) 0.20 (224) <0.01 (0.01) 0.48 (220) 0.01 (0.01) 0.66 (217)
 Internet versus email group –0.03 (0.02) –1.84 (220) –0.03 (0.01) –1.67 (217)
 Internet versus discussion group 0.03 (0.02) –1.84 (220) 0.02 (0.01) 1.60 (217)
 BMI <0.01 (0.01) 0.48 (220) <–0.01 (0.01) –0.06 (217)
 Age –0.02 (0.01) –2.22 (220)* –0.03 (0.01) –2.80 (217)
 Health reason 0.03 (0.01) 1.29 (217)
 Appearance to others –0.01 (0.01) –0.83 (217)
 Appearance to self 0.04 (0.01) 2.04 (217)*
–PRPE %: χ 2 (∆ df) 2.94% 9.89 (4) 3.02% 13.28 (3)*

SE: standard error; BMI: body mass index; PRPE: proportional reduction in preduction error.

Model 2 is compared to the unconditional model that included no predictors of the dependent variable.