|
Beginnings
The CROS
project started in 1997. Consumers and professionals were
beginning to adopt the recovery model. They wanted to
demonstrate achievements in rehabilitation that went beyond
symptom reduction and decreased levels of service usage. Our
mission was to create, with consumer and staff input, an
outcomes system that would provide meaningful data in a useful
way.
Questions were drafted based on feedback obtained through
focus groups of consumers and staff who were asked what
information would be useful in measuring consumers movement
toward their personal definition of recovery. Later, consumer
and staff focus groups addressed the clarity, ease and
accuracy of interpretation of the questions, the format of the
questionnaires, the burden and time needed for completion of
the questionnaires, and the format and content of the various
reports. The initial Consumer & Staff versions of CROS were
piloted and revised into CROS Version 1.0, which included a
draft VIP Questionnaire. After analyzing internal consistency
and scale correlations, CROS 2.0 was developed and fully
implemented in outpatient mental health settings. CROS 3.0 was
created in order to shorten the scales and provide congruence
between questions on the three separate questionnaires so that
scale scores could be directly comparable. A CROS training
course was designed and questionnaire and report processing
was automated.
Current Research
An
analysis of the psychometric characteristics of CROS 3.0 was
recently completed on a sample of 585 matched Consumer and
Staff CROS 3.0 questionnaires. The findings are summarized
below.
-
The most
common primary diagnosis (49%) was some type of schizophrenic
disorder; 20% had a primary diagnosis of depressive disorder;
17% had a primary diagnosis of bipolar disorder.
-
The
oblique factor analyses of the 35 items on the consumer
questionnaire and the 30 items on the staff questionnaire
indicate a satisfactory relationship to the five
conceptualized scales.
-
Consumer
and staff scale score correlations are moderately high. Scores among the consumer scales and
among the staff scales are substantially correlated with each
other.
-
The
reliability assessments of internal consistency and
test-retest reliability for both staff and consumer
questionnaires are satisfactory. Staff inter-rater reliability
measures were weaker but adequate and can be improved by
ongoing training. Concurrent validity was also established for
both the consumer and staff forms.
-
Changes in
scale scores over time were examined. Scale scores between the
first and second test administrations were significantly
correlated. All scores increased between the first and second
test administrations and were statistically significant in two
consumer scales and on all four staff scales.
-
Scores
on all nine scales were negatively skewed (loaded on the high
end of the scale), a common finding in many survey
instruments. Consumer scores are more skewed than staff scores
and consumer scale score means are significantly higher than
staff scale score means. Thorough orientation to the rationale
for the assessment and detailed instructions on completing the
questionnaires often decreases skewness.
This
psychometric analysis reveals a number of strengths; the
factorial structure is sound, consumer and staff scale scores
are significantly correlated with each other without being
redundant, measures of reliability are quite satisfactory and
initial measures of concurrent validity are encouraging and,
in spite of the low ceiling, significant improvements over
time were noted on six of the nine scales.
|
