The Journal of College and University Student Housing - Volume 43, No. 2 - 19

Development, Validity, and Reliability of the Campus Residential Experience Survey

factors to retain was influenced by analyzing
the results of eigenvalues, scree plot analysis,
total variance explained, and model fit.

Limitations
Before sharing results, it is important to
note limitations to this research. We took our
sample from the membership of ACUHO-I
because this association comprises student
affairs administrators who are arguably the
most invested in the residential experience of
colleges. However, ACUHO-I does not represent all student affairs administrators or all
residence life professionals. Therefore, some
administrators involved in either student
affairs or residence life, but not members of
ACUHO-I, were excluded from our study.
Using the ACUHO-I membership as our
sample should be taken into account when
generalizing the findings of this study. Participants were primarily from higher education
institutions in the United States.
Another limitation is the use of self-reported data. Our instrument measures perceptions of current status and desired status for
four latent variables. Self-reporting allows the
participant to honestly state a perception, but
response bias can influence self-reporting. To
address possible bias, all responses were anonymous. We offered no rewards or punishments
for responses of any type, and we offered no incentives in an attempt to boost response rates.

RESULTS
Our first research question sought to determine if there was construct validity (leading to
the measurement of unique latent variables)
in our survey instrument subscales. To evaluate construct validity, we conducted principal components analysis for our exploratory

factor analysis to determine if items did in fact
measure the latent variables of institutional
importance of students living on campus, the
role of residence life professionals as educators, faculty engagement in residential environments, and facility design of residential
environments. We conducted this analysis
twice - separately for the current status items
and the desired status items. Evaluations based
on eigenvalues, scree plot analysis, total variance explained, and model fit indicated an
eight-factor solution as most appropriate.
These eight factors corresponded to the eight
latent variables (subscales) of our instrument.
Therefore, we determined that all eight of our
theoretically-based scales had construct validity. All items loaded onto one of the factors, and
we did not need to remove any items from the
scale (see Table 2).
Our second research question sought to
determine evidence of reliability in the measuring of these constructs. To answer this
question, we conducted Cronbach's coefficient
alphas as a measure of internal consistency
and reliability for the eight scales that demonstrated construct validity (DeVellis, 2012).
These analyses were conducted separately for
the current status items and the desired status
items, and we intended to modify the scales
based on these analyses. For instance, if a particular item on a scale lowered the reliability of
the scale significantly, we would remove that
item. However, no items needed to be removed
due to issues with reliability. Using a threshold of .70 for deeming a scale reliable, all eight
scales were deemed to have sufficient reliability. Alphas ranged from .73 to .87. See Table 2
for the number of items in each scale and the
scale's corresponding alpha.

V O L U M E 4 3 , NO. 2 * 2 0 1 7

The Journal of College and University Student Housing - Volume 43, No. 2

Table of Contents for the Digital Edition of The Journal of College and University Student Housing - Volume 43, No. 2