Skip to main content
. Author manuscript; available in PMC: 2012 Oct 14.
Published in final edited form as: J Aging Health. 2010 Nov 10;23(3):454–480. doi: 10.1177/0898264310385930

Table 3.

Within-persons and Between-persons Differences in Body-Mass Index: Hierarchical Linear Modeling Results.

Parameter Variable M0 M1 M2 M3 M4 M5
Fixed Effect
Time-varying variables
Proxy −0.606*** −0.612*** −0.612*** −0.658*** −0.568***
Marital Status   0.256***   0.285***   0.210***
Assets −0.008   0.010   0.013
Income −0.095*** −0.078*** −0.059***
Self-Rated Health −0.006 −0.003
Index Disease   0.002   0.001
NAGI Index   0.241***   0.167***
CES-D Index −0.013 −0.019
Intercept π0
Intercept 27.383*** 27.613*** 27.370*** 27.290*** 27.361*** 27.517***
Mortality −0.436*** −0.531** −0.610*** −0.713*** −0.690***
Attrition −0.686*** −0.790*** −0.754*** −0.708** −0.197
Black   1.585***   1.446***   1.417***   0.145
Hispanic   0.687***   0.221   0.247 −0.144
Female −0.146 −0.159 −0.259** −0.083
Age (1992) −0.063*** −0.072*** −0.074*** −0.051***
Education −0.099*** −0.085*** −0.009
BMI (1992)   0.696***
Linear Slope π1
Intercept   0.047***   0.066***   0.073***   0.075***   0.067***   0.063***
Mortality −0.079*** −0.068*** −0.069*** −0.056*** −0.062***
Attrition −0.002   0.000   0.001   0.002 −0.018
Black −0.053*** −0.054*** −0.056*** −0.008
Hispanic −0.047*** −0.046*** −0.047*** −0.032*
Female   0.010   0.011   0.014   0.007
Age (1992) −0.006*** −0.005*** −0.005*** −0.006***
Education   0.000 −0.000 −0.002**
BMI (1992) −0.025***
Quadratic Slopea π2
Intercept   0.002**   0.001 −0.001 −0.001 −0.001 −0.001

Random Effect VARIANCE VARIANCE VARIANCE VARIANCE VARIANCE VARIANCE
Intercept1 20.054*** 19.916*** 19.510*** 19.372*** 18.191***   6.835***
Linear   0.074***   0.072***   0.072***   0.071***   0.070***   0.060***
Quadratic   0.001***   0.001***   0.001***   0.001***   0.001***   0.002***
Level1 R   7.579   7.558   7.558   7.553   7.525   7.114
AIC 445,993.33 445,607.26 445,411.33 445,326.31 444,765.75 424,447.38
*

p< .05,

**

p< .01,

***

p< .001.

a

Regression coefficients associated with all covariates on quadratic slope are non-significant in models 1 through 5.