Randall S. Davis
Assistant Professor
Southern Illinois University, Carbondale
MPA Program, Department of Political Science
Faner Hall, Mail Code 4501
Carbondale, Illinois 62901
E-mail: randalldavis3@gmail.com
Phone: 618-536-2371
Zachary Mohr
Assistant Professor
University of North Carolina Charlotte
Department of Political Science and Public Administration
E-mail: zachmohr@gmail.com
Phone: 316-613-9680
Paper presented at the 11th Public Management Research Conference, Madison, WI, June 20
22, 2013.
H1: Employees who perceive consistently applied rules will indicate higher job
satisfaction than those who perceive inconsistently applied rules.
H3: Employees who understand rules purposes will indicate higher job satisfaction than
employees who do not understand rule purposes.
Our theoretical constructs are measured using multiple survey items. First, formalization
is defined by three questionnaire items that assess the degree to which employees observe
organizational rules as available in writing. Second, three items assess the degree to which
organization members feel rules are optimally controlling. Third, three measures evaluate the
[INSERT TABLE 2]
Findings
Prior to examining relationships between constructs it is necessary to determine if the theoretical
model is an accurate representation of the data. General rules suggest that RMSEA .08 ,
CFI .90 , and NNFI .90 indicate models that adequately fit the data. The results from the
[INSERT FIGURE 1]
In the structural equation model we introduced four demographic covariates to rule out
possible alternative explanations. The inclusion of these covariates increased the model fit to
acceptable levels on all three fit statistics (see figure 2). The findings we present in the structural
equation model also suggest that the relationships between job satisfaction and elements of
formalization are complex. First, two of the three hypothesized relationships are significant.
Those employees who report more optimally controlling rules also report greater job satisfaction
( p .05 ). Likewise, employees who believe rules to be more consistently applied also report
higher degrees of job satisfaction ( p .001 ). Second, although we did not specifically
hypothesize about these effects, rule formalization significantly correlates with higher levels of
the other green tape attributes. In particular, employees who perceive higher levels of
formalization also believe that rules are more optimally controlling, perceive rules as more
consistently applied, and better comprehend rules. Figure 2 provides the standardized parameter
estimates from the structural equation model
Finally, we tested a model where rule formalization was used to predict job satisfaction.
That relationship was insignificant, suggesting that the influence of written rules on job
satisfaction is fully mediated by the other rule attributes. This finding comports with our
expectation that written rules increase job satisfaction through more optimal control, consistent
rule application, and rule comprehension. Although the SEM diagram illustrates several direct
relationships, it does not provide information of the total effect of written rules on job
satisfaction. Total indirect effects are the product of multiple direct effects (Kline, 2005). In our
analysis the indirect effect of written rules on job satisfaction through optimal rule control is .125
( p .05 ), the indirect effect of written rules on job satisfaction through consistent rule
application is .256 ( p .001 ), and the indirect effect of written rules on job satisfaction through
rule comprehension is .045 (n.s.). The total indirect effect of written rules on job satisfaction is
estimated as the sum of all indirect effects, which in this model is .426 ( p .001 ).
Unlike traditional regression models that estimate a single R 2 value associated with the
dependent variable, structural equation models estimate multiple R 2 values for each endogenous
variable. First, the demographic covariates in the model explain 1.6% of the variation in written
rules. Second, the demographic covariates and the written nature of rules explain 35.2% of the
variation in optimal rule control, 73.9% of the variation in consistent rule application, and 72.5%
of the variation in rule comprehension. Finally, the three rule attributes and demographic
controls explain 32.9% of the variation in job satisfaction. Taken as a whole the R 2 values
suggest that our model has reasonable explanatory capacity.
Finally, there are a few significant relationships between demographic control variables
and other model constructs. First, female survey respondents indicate working in less formalized
[INSERT TABLE 4]
Percent
Gender
Male
Female
Missing
948
698
19
56.9
41.9
1.1
1215
120
243
57
73.0
7.2
14.6
3.4
30
1.8
39
70
2.3
4.2
87
5.2
320
388
224
230
307
19.2
23.3
13.5
13.8
18.4
Mean
S.D.
45.33
10.56
10.555
14
Race
White
Hispanic
Black
Other (e.g., Asian,
Pacific Islander, etc.)
Missing
Organizational Role
Department head
Division
head/superintendent
Administrative or
policy staff
Supervisor/manager
Lead worker
Clerical
Technical
Missing
Age
% Missing
Minimum
Maximum
Mean
Std.
Deviation
Written1
1467
11.89%
3.7519
1.14134
Written2
1298
22.04%
3.3598
1.0693
Written3
1259
24.38%
2.278
0.69331
Comprehension1
1456
12.55%
3.456
1.0977
Comprehension2
1277
23.30%
3.5309
0.94522
Comprehension3
1471
11.65%
3.4405
1.11409
Comprehension4
1291
22.46%
3.897
0.87395
Consistent1
1468
11.83%
3.2084
1.26788
Consistent2
1491
10.45%
3.053
1.18231
Consistent3
1471
11.65%
2.1407
0.73673
Control1
1448
13.03%
3.1222
1.13661
Control2
1467
11.89%
3.8228
1.10435
Control3
878
47.27%
3.3576
1.09508
Job Satisfaction1
1400
15.92%
3.8721
0.95207
Job Satisfaction2
1398
16.04%
3.922
0.99767
Job Satisfaction3
1395
16.22%
3.8186
0.95114
.420
.470
.826
.837
.860
.581
1.0
.657
Control
1.0
Written
.511
.831
Consistent
1.0
1.0
.500
Comprehension
Job Sat.
1.0
Model Fit: (88, n=1,523) = 1256.829, p < .001; RMSEA = .093(.089, .098); CFI = .961; NNFI(TLI) = .947
Control
.220*
.570***
.375***
.833***
Written
.846***
Consistent
.302***
.377***
.834***
.054
Comprehension
Model Fit: (133, n=1,313) = 1195.671, p < .001; RMSEA = .078(.074, .082); CFI = .962;
NNFI(TLI) = .948
Note: * p .05 ; ** p .01 ; *** p .001
Job Sat.
Variable
Age
Female
Nonwhite
Managerial Role
Written Rules
Estimate
S.E.
-0.003
0.002
-0.091
0.041
0.102
0.045
0.037
0.042
Est./S.E.
-1.593
-2.216
2.253
0.889
p
0.111
0.027
0.024
0.374
Variable
Age
Female
Nonwhite
Managerial Role
Optimal Control
Estimate
S.E.
0.001
0.001
0.040
0.019
-0.008
0.020
0.082
0.021
Est./S.E.
0.622
2.185
-0.416
3.968
p
0.534
0.029
0.678
0.000
p
0.001
0.788
0.683
0.000
Variable
Age
Female
Nonwhite
Managerial Role
Rule Comprehension
Estimate
S.E.
Est./S.E.
0.002
0.002
0.984
0.106
0.043
2.487
-0.026
0.047
-0.550
0.266
0.045
5.904
p
0.325
0.013
0.582
0.000
Variable
Age
Female
Nonwhite
Managerial Role
Job Satisfaction
Estimate
S.E.
0.011
0.002
0.050
0.051
0.161
0.056
0.117
0.051
p
0.000
0.327
0.004
0.021
Est./S.E.
4.747
0.981
2.888
2.306