teams,
which enhances perceptions of control (e.g., Thomas & McDaniel,
1990) and feasibility. The most relevant organizational-level
factor is the availability of resources, which is at the core
of this conflict, and has been found to affect perceptions of
control and feasibility (Dutton, Stumpf & Wagner, 1990).
METHOD
This
study was conducted as part of a larger project funded by the
Ohio Department of Mental Health, Office of Program Evaluation
and Research (grant #91-1003), examining the impact of the Act
on Boards and mental health agencies. The perceptual measures
used here were assessed in a mail survey of 540 decision makers
in Ohio's 53 Boards. The work experience measure was a composite
of age, managerial experience and total work experience, averaged
across respondents within Boards. Information processing capacity
was measured by Thomas and McDaniel's (1990) 9-item scale. Resource
adequacy, both currently and in the future, was assessed with
an 18-item scale. Control frame was measured by a 5-item scale,
perceived feasibility of implementing the Act by an 8-item scale,
and the quality of the working relationship with ODMH by a 4-item
scale. Board stance in the lawsuit, obtained from lawyers involved
in the case, was coded so that higher numbers indicate greater
resistance: defendant = 1, intervenor = 2, non-party = 3, plaintiff
= 4. The organization was the referent for all perceptual items
and was the focal unit for hypothesis testing.
RESULTS
The
individual response rate for the survey was 29% (156/540), but
the rate was better for higher-level individuals (e.g., 68% for
Executive Directors). The response rate for Boards was 91% (48/53),
with an average of 3.3 responses per Board. The internal consistency
of the scales, assessed at the individual level, was acceptable;
Cronbach alphas ranged from .74 to .91. Convergence among respondents
within Boards was assessed with James, Demaree & Wolf's (1993)
IRR index, except for the experience measure, which was not expected
to converge across individuals. The IRR values for the perceptual
measures (.77 to .91) justify aggregation.
Separate
path analyses were conducted for the models with feasibility versus
control, using RAMONA, a structural equation modeling program
contained in the SYSTAT statistical package (Wilkinson & Hill,
1994). Model testing followed Medsker, Williams and Holahan's
(1994) guidelines. The a priori models and subsequent revised
models were compared to a null model with no causal links. Four
fit indices were used: the normed and nonnormed fit indices (NFI,
NNFI), the comparative fit index (CFI), which is resistant to
errors associated with sample size, and the parsimonious model
fit index (PFI), which adjusts for degrees of freedom in the target
model. Values of .90 and greater indicate adequate fit for the
first three, and .60 and above has been suggested as the guideline
for PFI.
The
a priori model using control did not show adequate fit: NFI =
.74, NNFI = .72, CFI = .84, PFI = .49. Comparison of the original
and reproduced correlations suggested three additional direct
paths, as described below. This slightly revised model demonstrated
excellent fit: NFI = .94, NNFI = 1.13, CFI = .95. PFI (.55) is
somewhat low, but is deflated by the modest sample size.
The
results for the revised model using control (Figure 1) support
four of the five predicted paths. Quality of working relationship
has a strong negative path to organizational resistance, and the
path from control to quality of working relationship is positive.
(The lack of a direct path from control to organizational resistance
suggests that quality of working relationship is a complete mediator
here.) Two context variables (work experience and resource availability)
show direct paths to control. Three paths were added: from experience
to working relationship and to organizational resistance, and
from information processing capacity to organizational resistance.
The
a priori model using feasibility did not show adequate fit: NFI
= .65, NNFI = .50, CFI = .71, PFI = .38. Comparison of the original
and reproduced correlations suggested four additional paths, as
described below.
Research
Home - Page 1 | 2
| 3 | 4 | 5
| <PREV NEXT
>
