riptive Analysis
In this and the following sections we provide
descriptive analyses of the sleep and outcome data
followed by findings derived from the cross-domain
latent growth modeling. Table 1 presents the means,
standard deviations, and univariate skewness and
kurtosis coefficients for sleep and the outcome
variables. The variables were normally distributed
with no significant skewness or kurtosis; skewness
coefficients ranged from .52 to 1.54, whereas kurtosis
Sleepless in Chicago 87
ranged from – 0.51 to 2.65. As predicted, overall
observed means showed declines in hours of sleep,
self-esteem, and self-reported grades, and a rise in
depressive symptoms. The pattern of associations
between the predictors and the outcomes indicated
significant correlations among the variables in the
expected direction (see Table 2). Hours of sleep in
eighth grade were negatively correlated with depressive
symptoms in eighth grade and positively
correlated with self-esteem in eighth grade, such that
students obtaining less sleep were more likely to
report depressive symptoms and low self-esteem.
There was a positive correlation between eighthgrade
sleep and self-reported grades, such that
students obtaining more sleep reported slightly
higher grades.