Journal of College & University Student Housing Volume 48, No. 2 - 100

Relationships Matter
Data Analysis and Variables
For each analysis, all responses with missing data were excluded under pairwise deletion
parameters (Peugh & Enders, 2004). Descriptive analysis provided an overview of
the sample. We then conducted three separate multiple linear regression analyses with
each dimension of burnout (emotional exhaustion, depersonalization, and personal
accomplishment) serving as a dependent variable. The independent variables used in
the regression analyses included RAs' gender, race/ethnicity, classification, number
of residents for which they were directly responsible, residential area (apartments versus
residence halls), and total mattering. Due to lack of reported gender identification
outside the male-female binary, gender was coded dichotomously. Race and ethnicity
(White versus underrepresented race/ethnicities) were coded dichotomously. Classification
of the RAs originally included four categories: sophomore, junior, senior, and
graduate students. Due to the graduate student classification containing only two RAs,
we analyzed the data with graduate students combined with seniors and individually.
Since there was no meaningful difference in the results, we proceeded with the combined
classification of Senior+ to reduce the number of groups to three. The type of
residential areas was also a dichotomous variable (apartments versus residence halls).
By virtue of Cronbach alpha levels exceeding 0.7 (DeVellis, 2003; Kline, 2005),
each dimension of burnout was deemed internally consistent: emotional exhaustion
(α = .928, n = 9); depersonalization (α = .737, n = 5); and personal accomplishment
(α = .887, n = 8). The original researchers provided specific instruction to consider each
burnout dimension independent of the others and not combined (Maslach et al., 1996).
Each component of mattering was also deemed consistent: awareness (α = .815,
n = 8); importance (α = .839, n = 7); ego extension (α = .737, n = 3); and reliance (α = .843,
n = 6). Unlike the burnout instrument instruction, no such guidance was articulated
for the necessity of treating mattering subscales independently; therefore, due to the
high correlations among the mattering subscales, we summed the individual component
scores to create a total mattering composite score for use in the regression models
(α = .938, n = 24).
LIMITATIONS
The lack of diversity in the sample (specifically, ethnicity, gender, and regional affiliation)
should be cautiously considered when generalizing the results to other populations.
Additionally, the dichotomization of variables unfortunately limits the voices
of many RAs from underrepresented groups. Limitations also exist with the modified
UMS instrument and the associated wording of particular items. For example, the
item labeled " Quite a few people in my community look to me for advice on issues
of importance " may elicit a different interpretation from participants on what exactly
" quite a few people " entails. Additionally, some adverbs used in the instrument may
be deemed ambiguous, as demonstrated by usage of the word " sometimes " in the
statement " Sometimes in my community, I feel as if I were invisible. " Finally, the study
100 The Journal of College and University Student Housing
Journal of College & University Student Housing Volume 48, No. 2

Table of Contents for the Digital Edition of Journal of College & University Student Housing Volume 48, No. 2
