Professional Documents
Culture Documents
Anderson1995 (D)
Anderson1995 (D)
Volume 26 Number 5
SeptJOct. 1995
Printed in the i%S.A.
ABSTRACT
Despite the impact that Deming and his 14 Points have had on the practice of quality
management, empirical support for the effectiveness of the Deming Management
Method has not advanced beyond the presentation of anecdotal, case-study evidence.
In part, this is because theory to guide the conduct of empirical research has not
been available. Only recently has such a theory of quality management to describe
and explain the effectiveness of the Deming Management Method been articulated
in the literature.
This paper continues the journey of theory development; it reports the results
of an exploratory empirical analysis of an articulated theory of quality management
underlying the Deming Management Method. The constructs in the proposed theory
are operationalized using measurement statements developed by the World-Class
Manufacturing research project team at the University of Minnesota and Iowa State
University. Path analysis is applied to the World-Class Manufacturing project data
to explore the empirical strength of relationships advanced in the theory. The path
analytic: results provide support for several of the proposed relationships in the
theory, and more importantly, suggest a number of new relationships which have
not heretofore been proposed.
Subject Areas: Deming Management Method, Path Analysis, Quality Management, Theory
Development, and Theory Testing.
*An earlier version of this paper was presented at the 1993 Annual Meeting of the Decision
Sciences Institute in Washington, D.C.
637
638 A Path Analytic Model
INTRODUCTION
Quality management has increasingly become the focus of organizations competing
in both domestic and global economies. Quality has often been cited as the highest
competitive priority (e.g., [29]), an issue of strategic importance and survival (e.g.,
[181), and a means of competitive performance (e.g., [4]). Numerous descriptive
case studies of quality and other organizational improvements that accompany the
adoption of quality management approaches have found their way into the literature
(e.g., [16] [35]). Theorizinghow and why this occurs,and testing this theory empirically,
beyond case-based research, however, are far less common.
A case in point is W. Edwards Deming and his Deming Management Method
[7] [8] [9] [lo]. The Deming Management Method contains a set of 14 Points,
presented by Deming to be “principles of transformation” to improve the practice
of management [6, p. 231. Deming’s 14 Points have been embraced by many organiza-
tions, both domestically and internationally, across manufacturing and service
industries and profit and nonprofit sectors.Many firms (including Ford Motor Company,
Xerox Corporation, Zytec Corporation, etc.), have attributed the turnaround of their
organizations,in varying degrees, to the Deming Management Methd, their “success”
stories have been documented in Baker and Artinian [2], Hodgson [19], Scherkenbach
[30], and Walton [35, pp. 121-2381,
Despite the impact that Deming and his 14 Points have had on the practice of
quality management, empirical support for the effectiveness of the Deming Manage-
ment Method has not advanced beyond the presentation of anecdotal, case-study
evidence. In part, this is because theory to guide the conduct of empirical research
has not been available. Only recently has such a theory of quality management to
describe and explain the effectiveness of the Deming Management Method been
articulated by Anderson, Rungtusanatham, and Schroeder [l].
In this paper, the proposed theory of quality management underlying the Deming
Management Method is empirically examined. The constructs in this theory are
operationalized using measurement statements developed by the World-Class Manu-
facturing research project team at the University of Minnesota and Iowa State
University. Path analysis is applied to the World-Class Manufacturhg project data
to explore the empirical strength of the relationships advanced in the theory.
It is important to note that this paper does not claim to present formal conclusive
tests of the proposed theory; the secondary nature of both the construct operation-
alization and the data limits the ability to do so. The reported results are, therefore,
intended to be preliminary empirical observations conducive to further theory develop-
ment, marking a continuation of a journey begun by the original architects of the
proposed theory.
Because this paper marks a continuation of the work of Anderson, et al. [l], the
intent is not to reiterate further the details of the processes of theory development
and theory juxtaposition in literature, both of which have been fully described (see
[11). Several additional comments about the proposed theory are, however, appropriate
for the purposes of this study.
First, the articulated theory of quality management, quite obviously, has its
foundations in Deming’s 14 Points and other writings. The proposed constructs and
relationships, while stemming from the Deming Management Method, are also
reflected to varying extent in the writings of other quality advocates, as well as in
various national quality awards. The importance of leadership, for example, is high-
lighted by Crosby [6] and Juran [20], and in the Malcolm Baldrige National Quality
Award and the Deming Prize (see discussion in Dooley, Bush, Anderson, and
Rungtusanatham [ 111). The tools and techniques used in process management are
another similarity between Deming and Juran. Other similarities and differences can
be gleaned from Garvin [17].
A second comment concerns the generalizabilityof the proposed constructs and
relationships. As pointed out in Anderson, et al. [l], the Deming Management
Method has been implemented worldwide (see Hodgson [ 19]), suggesting an ease
of transferability of the prescriptions in the 14 Points, as well as the constructs and
relationships proposed in the path diagram in Figure 1. Conversely, others (e.g., [39])
Anderson, Rungtusanatham, Schroeder, and Devaraj 641
Couperution
Lfll Iniprovenient
I’rucess
Orgaaizaliunal
Syrlciri
I Pullillnient
I
I
I 1
i i I I I
I
I
I
! ! t i 1
I
1
I
I
I
I
have argued that there may be cultural constraints on the applicability of the Deming
Management Method across different national cultures. One approach to resolving
this debate is by conducting an empirical investigation comparing the level and
success of implementing the Deming Management Method andor assessing the
validity of the proposed theory of quality management underlying the Deming
Management ]Method across national cultures. While the data is not available for
unaerraKlng siicn a sruay at tnis point, me empincai sway reponeu in mis paper
marks a beginning towards addressing this issue by exploring the validity of the
articulated theory using both American-owned and Japaneseamed plants in the U.S.
As noted earlier, this paper complements and furthers the theory development
task by subjecting the articulated theory to an empirical analysis, for the process of
theory development demands that any newly-proposed theory be confronted with
empirical realiig so as to substantiate and/or enhance a theory’s descriptive, explanatory,
and prescriptive properties.
To conduct this empirical examination, three questions need to be addressed in
a satisfactory manner:
1. How can the constructs and conceptual relationships be operationalized
for empirical study?
2. What are the empirical strengths of the relationships specified in the
proposed theory?
3. Are other relationships observable from an empirical standpoint?
The answers to these questions provide a reality-based foundation for modifying
and refining the proposed theory, making it more descriptive and prescriptive of the
effectiveness of the Deming Management Method.
642 A Path Analytic Model
j= 1
Anderson, Rurrgtusanatham, Schroeder, and Devaraj 643
where
Lick) = plant-level score for the ith measurement item in the kth measurement
scale,
IV = jth individual respondent score for the ith measurement item,
ni = number of respondents for ith measurement item.
A score for the kth measurement scale at the plant level (Sk)is then computed by
averaging the: plant-level individual item scores for the set of items defining the kth
measurement scale (see Equation 2):
where
Sk = plant-level score for the kth measurement scale,
Ii(k)= ith measurement item for the kth measurement scale,
ikfk = number of measurement items making up the kth measurement scale.
lbble 2: (continued).
Employee Fulfilment (a= 0.6438)
I would feel unhappy if I could not take pride in my job.
Doing a good job should mean as much to a worker as a good paycheck.
If I do a sloppy job at work, I feel a little ashamed of myself.
Only plants with more than 100 employees from the three pre-specified industries
were retained and included in the population. An equal number of plants was randomly
selected from each of the nine master lists to total 72 plants. Forty-one out of 72
plants chose to participate, corresponding to a 60 percent participation rate. Table 3
shows the distribution of plants within each strata.
Plant managers andor vice presidents of manufacturing of target plants were
contacted by telephone and by letter to generate interest and obtain approval for
conducting the study. Each plant that agreed to the study was asked to assign a
coordinator to facilitate and handle survey administration. Q U e s t i ~ ~ a iwere
r e ~ admin-
istered to individuals with different job titles and responsibilities in the plant: plant
manager, accountant, production and inventory manager, quality manager, process
engineer, research coordinator, supervisors, and workers. Where multiple individuals
were available for a particular job class, the coordinator randomly selected a pre-
specified number of informants from within the job class.
646 A Path Analytic Model
Disturbance
Coefficients p-value Variable $ Term
0.59 o.ooo1 A -N- -N-
0.46 0.002 B 0.35 0.81
0.71 o.ooo9 C 0.21 0.89
-0.17 0.4 D 0.35 0.81
0.21 0.002 E 0.21 0.89
0.27 0.085 F 0.07 0.96
0.15 0.317 G 0.18 0.91
0.35 0.026
The path diagram for the theory proposed in Figure 1 is shown in Figure 2.
According to the proposed theory, Visionary Leadership is posited to exert simulta-
neous, direct, and linear influences on both Internal and External Cooperation and
Learning. The path diagram in Figure 2, therefore, shows two paths emanating from
Visionary Leadership, with one path leading to Internal and External Cooperation
and the other leading into Learning. The strength of the Visionary Leadership’s
influence is deinoted by the value of the path coefficient associated with each path;
in the case of the Visionary Leadership + Internal and External Cooperation relation-
ship, the magnitude of the proposed relationship is captured by the path coefficient,
PBA-
Likewise, both Internal and External Cooperation and Leaming are postulated
as affecting Prccess Management, which, in turn, simultaneously affects Continuous
Improvement and Employee Fulfillment-accounting for paths pDB, pDc, pED and
pFD.Figure 2 also shows paths pGE andpGFto operationalize the effect of Continuous
Improvement and Employee Fulfillment on Customer Satisfaction.
Mathematically, path analysis decomposes the empirical correlations or covariances
among the measured variables to estimate the path coefficients in the path diagram
[27]. Mechanically, the path coefficients can be estimated in a number of ways (e.g.,
tracing rules, the First Rule, etc.) [21]. Most empirical applications of this methodology
have employed multiple regression analyses (i.e., ordinary least squares estimation).
It can be shown that the path Coefficients are identical to the standardized regression
coefficients [38:1 (see Li [24] for proof). Therefore, to estimate pBA one would regress
Internal and External Cooperation on Visionary Leadership.
648 A Path Analytic Model
Paralleling multiple repssion analysis, path analysis uses the error term (distur-
bance term), to capture the effects of all other variables not explicitly captured in
the path diagram. The path analytical results can be interpreted in a manner consistent
with the evaluation of multiple regression results. Tests of the significance of path
coefficients and the overall significance of individual relationships can be examined
by conducting t-tests and by examining coefficients of determination,R2s, respectively.
The t-test indicates whether or not a path coefficient differs statistically from zero,
that is, whether or not the hypothesized linear relationship holds. The coefficient of
determination, on the other hand, indicates the amount of variance in the dependent
variable which is accounted for by the variables entered in the path or structuralequation.
TransportatioriComponents
A Visionary Leadership 0.05
B Internal 8t External Cooperation 0.05
C Learning 0.05
D Process Management 0.10
E Continuoius Improvement -
F Employee Fulfilment 0.05
G Customer Satisfaction 0.005
0.005
Transportation Components
A Visionary Leadership -
B Internal & External Cooperation --
C Learning -
D Process Management -
E Continuous Improvement 0.05
F Employee Fulfillment -
G Customer Satisfaction -
ap<o.001
bp<O.Ol
'pcO.05
Notes:
A = Visionary Leadership
B = Internal & External Cooperation
C = Learning
D = ProcessManagement
E = Continuous Improvement
F = Employee Fulfillment
G = Customer Satisfaction
Anderson, Rungytusanatham, Schroeder, and Devaraj 65 1
covariance between any two variables into three components: direct (D), indirect
(N), and unexplained (U) effects [22].
A direct effect is posited to exist if a single arrow connects two variables. When
two variables of interest are connected only via other intervening variables-in other
words, the relationship is represented by compound paths with no direct path between
the two variablles of interest-the effect is deemed to be indirect. In the context of
Figure 2, Visionary Leadership has a direct effect on Internal and External Cooperation,
but has an indirect effect on Process Management through its influence on Internal
and External Cooperation and Learning. The total indirect effect, which can be
traced along all1 possible compound paths connecting two variables, is computed by
summing the values of all possible compound paths, where a compound path value
is simply the product of the path coefficients constituting the compound path. In
this case, the total indirect effect (NDA) of Visionary Leadership on Process Man-
agement is given by the following expression:
where
rji = ernpirical correlation between variables i andj;
Uji = portion of the empirical correlationbetween variables i andj not accounted
for by the sum of the direct and indirect effects of i on j ;
Dji = direct effect of variable i on variablej, denoted by the value of the path
coefficient Pji;
N i = total indirect effect of variable i on variablej.
For the path diagram in Figure 2, the computed direct, indirect, and unexplained
effects are shown in Table 7.
According to Figure 2, Visionary Leadership is posited to only have indirect
effects on F’rocess Management, Continuous Improvement, Employee Fulfillment,
and Customer Satisfaction. After factoring out these indirect effects, there remains
considerable residual unexplained effects, suggesting that there may be direct effects
of Visionary Leadership that need to be accounted for. For example, the unexplained
effect of Visionary Leadership on Continuous Improvement is 0.19; 55 percent of
652 A Path Analytic Model
Notes:
A = Visionarykadership
B = Internal & External Cooperation
C = Learning
D = ProcessManagement
E = Continuous Improvement
F = Employee Fulfillment
G = Customer Satisfaction
DISCUSSION OF RESULTS
The path analytic results, in general, indicate support for many of the relationships in
the proposed theory. Ofthe eight direct paths, only two are statisticallynon-significant-
the relationship between Learning and Process Management and between Continu-
ous Improvement and Customer Satisfaction. In the first case, the failure to
statistically uncover a significant relationship may be attributed to the multicol-
linearity problem explained earlier. The latter case, on the contrary, is intriguing and
counter-intuitive because, conceptually, Continuous Improvement and Customer
Satisfaction should be related. Several explanationscan be offered for this surprising
result: the measurement of Continuous Improvement may be problematic, the sec-
ondary nature (of the data may have affected the outcome, or the aggregation of
individual percieptions into a plant-level response may have reduced the magnitude
of the true relationship. In any case, it is not possible, given the purpose and nature
of this study, to explore this further.
While the remaining relationships are statistically significant, the coefficients
of determination, or $s, for some of the proposed relationships are rather low. A case
in point is the impact of Process Management on Continuous Improvement and
Employee FulfiHlment. It appears that Process Management, by itself, does not explain
654 A Path Analytic Model
a large portion of the observed variance. Perhaps there are other variables, either
captured in or excluded from the proposed theory, which may add considerable
explanatory power to the observed variances of Continuous Improvementand Employee
Fulfillment.
One statistically significant relationship worth further investigation is the direct
effect of Employee Fulfillment on Customer Satisfaction. The statistical significance
of this result, however, needs to be interpreted carefully because Customer Satisfaction
has been operationalized based on employee perceptions. In any case, the results
need to be examined and replicated with externally based measures of Customer
Satisfaction, in light of the increasing interest on the nature and strength of this
relationship. The literature, for example, is just beginning to examine the role that
satisfied employees play in ensuring customer satisfaction. While Deming did not
explicitly address this relationship, it is implicitly treated in the Deming Management
Method as a natural outcome of employees’being able to attain pride of workmanship.
The logic of this relationship in the context of Deming’s 14 Points quality manage-
ment philosophy rests fundamentally on the interaction of pride (of quality) of work,
and Deming’s enunciation of the three comers of quality [6, p. 1771 [7, p. 2281. As
explained by Anderson, et al.:
Pride of quality of work necessarily relies upon (a) understanding and satisfying
changing customers requirements, (b) providing relevant product- or service-related
information, and (c) anticipating customer usage. [I, p. 4961
The presence of large unexplained effects for the path model representation of
the proposed theory, nevertheless, suggests that the present theory may not be
exhaustive in its specification of plausible causal and correlational relationships. A
number of other plausible relationships can be suggested. For example, it is quite
likely that Visionary Leadership, in addition to indirectly affecting Process Manage-
ment through the creation of a highly cooperative and leaming environment, may
directly affect Process Management through leadership actions and policies embracing
Process Management practices. The empirical correlations reported in Table 6 cer-
tainly advocate a direct path from Visionary Leadership to Process Management, as
well as the existence of several other direct paths emanating not only from Visionary
Leadership but also from Internal and External Cooperation, Learning, Process
Management, and Continuous Improvement. The possibility of direct paths from
Visionary Leadership, Internal and External Cooperation, and Learning to both
Continuous Improvement and Employee Fulfillment is consistent with earlier insights
that other constructs, besides Process Management, may affect Continuous Improve-
ment and Employee Fulfillment. Similarly, Customer Satisfaction may be directly
affected by Visionary Leadership and Process Management.
Additionally, at least two bi-directional (i.e., correlational) relationships are
supported by the data. The multicollinearity problem, noted earlier, can be resolved
by drawing a bi-directional arrow linking Internal and External Cooperation and
Learning. Alternatively, one can specify a reciprocal relationship between the two
constructs, indicating a “looping” causal effect between the two. That is, Internal
and External Cooperation facilitates Learning, which, in turn, facilitates greater
Internal and External Cooperation. Of course, these changes should be made to the
proposed theory only if they can be supported theoretically.
Anderson, Rungtusanatham, Schroeder, and Devaraj 655
CONCLUSIONS
The empirical results reported here are based on an analysis of an available data
source. Becausle both Japanese-owned and Amencan-owned plants are included in
the analysis, and knowing Deming’s influence on Japanese management, the results
of this study provide a glimpse of the cross-cultural validity of the proposed theoretical
relationships. A more valid approach, of course, would be to empirically examine
the relationships in the proposed theory in plants located in multiple countries, a
task which is the focus of future research plans related to the proposed theory shown
in Figure 1.
More importantly, it must be emphasized that these results represent an inter-
mediate step in theory development, and more specifically, provide an opportunity
for theory refinement and reformulation. The limitations presented by the secondary
nature of the construct operationalization and data source, however, discourage the
interpretation of these empirical findings as formal statistical tests of the relationships
in the proposed theory. Nonetheless, the insights generated by analyzing secondary
656 A Path Analytic Model
data in the context of the proposed theory, besides lending credence to many of the
proposed relationships, also provides an opportunity to explore at least two addi-
tional relationships not specified in the original theory.
From an academic perspective, therefore, the empirical analysis of the proposed
theory has served to focus attention on previously unidentified and unconceptualized
relationships among the constructs in the proposed theory-relationships deserving
of further theoretical and empirical examination. Subsequent research needs to engage
in the development of more valid and reliable operational definitions for the pro-
posed constructs, overcoming the limitations posed by the data source used in the
present study. Once the proposed constructs have been operationally defined in a
valid and reliable manner, formal theory testing can proceed, with such results
contributing to the evolving scientific knowledge base on the Deming Management
Method and its organizational consequences and implications.
Even though the results are exploratory, there appears to be some practical
utility. From a pragmatic perspective, for instance, the empirical results suggest the
importance of organizational leadership and its impact on creating an organizational
form and instituting organizational practices to bolster the goal of organizational
survival. The results also suggest that while organizations can theoretically survive
by continually satisfying customers through innovations that lead to quality improve-
ments in processes, products, and services, efforts to enable organizational constituents
to derive happiness, satisfaction, and pride of work are potentially instrumental in
improved customer satisfaction as well-a conclusion that is intuitively appealing.
Lastly, the fact that many of the relationships appear to receive exploratory empirical
support should help organizations better understand and more effectively embrace
the adoption of quality management within organizations. [Received: October 5,
1994. Accepted: June 1, 1995.1
REFERENCES
Anderson, J. C., Rungtusanatham, M., & Schroeder, R. G. A theory of quality
management underlying the Deming Management Method. Academy of Man-
agement Review, 1994, 19(3), 472-509.
Baker, E. M., & Artinian, H. L. The case of Windsor Export Supply. Quality
Progress, 1985, 17(6), 61-69.
Bates, K. A., Amundson, S. D., Schroeder, R. G., & Moms, W. T. The crucial
interrelationship between manufacturing strategy and organizational culture.
Management Science, forthcoming.
Buzzell, R., & Gale, B. The PIMS principles-finking strategy to pelformance.
New York The Free Press, 1987.
Cronbach, L. J. Coefficient alpha and the internal shuchue of tests. Psychometrika,
1951, 16(2),297-334.
Crosby, P. Quality isfree. New York: Mentor Publishing, 1979.
Deming, W. E. Out of the crisis. Cambridge, MA: Massachusetts Institute of
Technology, Center for Advanced Engineering Study, 1986.
Deming, W. E. Quality, productivity, and competitive position. Cambridge:
Massachusetts Institute of Technology, Center for Advanced Engineering Study,
1982.
Anderson, Rungtusanat?tam,Schroeder, and Devaraj 657