Ethnic Identity and Adverse Mood Related Symptoms
Multiple regression and logistic regression were used to test the second
hypothesis: There is a relationship between ethnic identity and mental health functioning
in minority adolescents. The assumptions for linear regression include residual
homoscedasticity (i.e., equal variance of the dependent variable at each level of the
independent variable), normal distribution, and linearity. Also, all pertinent variables
must be included, and all irrelevant variables are excluded. Lastly, measures must be
reliable and observations must be independent (Licht, 1997). These assumptions were
met with the exception of linearity. Transformation of the data and/or use of a non-linear
model were not appropriate due to the large number of zero scores. However, multiple
regression is robust to the effects of violations of the error assumptions (Licht, 1997).
Table 4-7. Multigroup Ethnic Identity Measure by Ethnic Group
a Mean SD
African American .74 3.55 .61
European American .75 3.33 .53
Hispanic .83 3.47 .59
Other .85 3.59 .67
Three subscales of the MAYSI-2 (angry/irritable mood, depressed/anxious mood,
and somatic complaints due to anxiety) were summed to create one dependent variable (a
= 0.86). This variable was normally distributed. This sub-divides the general construct of
adverse mental health symptoms and creates two more specific constructs. The constructs
are mood related adverse mental health symptoms and substance use. The alcohol/drug
use measure was not normally distributed and this measure was analyzed separately
using logistic regression. The rationale for this separation is detailed in the following
section.