Description
1. The data (HEALTH.xlsx) are from a randomized, controlled trial among women of
childbearing age to evaluate the effects of an educational intervention. One response
variable of interest is the participants self-rating of health status as either good or
poor. The researchers would like to assess the effect of the intervention on self-rated
health across the follow-up period, as well as whether these effects are influenced
by the mothers age. There are n = 80 women enrolled in this trial. These data
were measured at 4 points in time: randomization, 3 months, 6 months, and 12
months post-randomization. Each observation in the dataset contains values for the
following variables (in this order):
ID: unique participant identification code
TIME: the visit number for this observation of this participant
1: corresponds to a participants visit at the time of randomization
2: corresponds to the 3 month post-randomization visit
3: corresponds to the 6 month post-randomization visit
4: corresponds to the 12 month post-randomization visit
TRT: the treatment group to which the participant has been randomized
Control
Intervention
HEALTH: participants self-rated level of health for this visit
Good
Poor
AGEGROUP: participants age group at time of randomization
15-24 (years old)
25 to 34 (years old)
35+ (years old)
(a) Evaluate the bivariate, cross-sectional relationship between randomized group
assignment and participants health self-rating at the time of randomization.
Interpret and discuss these findings.
(b) Perform a longitudinal data analysis across all study follow-up visits (but not
at randomization) to describe the relationship of the participants self-ratings
as a function of the effects of health self-rating at the baseline, treatment
group, month post randomization, and age group as predictors. Fit a GEE
with unstructured correlation structure. Interpret your results.
(c) Fit a generalized linear mixed effects model with subject-specific random intercepts. Interpret your estimates. How are the interpretations different from
the GEE model?
1 P8131
