Technovation 20 (2000) 539–550

www.elsevier.com/locate/technovation

The adoption of continuous improvement and innovation strategies

in Australian manufacturing firms
a,* b
Milé Terziovski , Amrik S. Sohal
a
Euro–Australian Centre for Global Innovation Management, Faculty of Business and Economics, Monash University, PO Box 197, Caulfield
East, VIC 3145, Australia
b
Quality Management Research Unit, Faculty of Business and Economics, Monash University, PO Box 197, Caulfield East, VIC 3145,
Australia

Received 12 August 1999; accepted 24 September 1999

Abstract

The purpose of this study was to investigate the adoption of Continuous Improvement (CI) strategies of a large random sample
of Australian manufacturing firms. The study was undertaken as part of a wider international survey investigating continuous
improvement practices in Australia, Denmark, Finland, Sweden, the Netherlands, and the UK. The survey was mailed to 1200
managers responsible for manufacturing organisations in Australia. A response rate of 32 per cent was obtained. The quantitative
data was analysed using a Statistical Package for Social Sciences (SPSS). The data analysis revealed that the motivation to adopt
CI was related to improved quality conformance, increased productivity, reduced costs, and improvement in delivery reliability.
Past experiences of CI were positively correlated with the length of time the process had been in use; the breadth of its application;
the percentage of employees actively involved in the program (for operators and non-operators) and training in problem solving.
Therefore, the critical implication for managers is that future management development initiatives need to include strategies to
assist managers with their understanding of the potential benefits of the CI process, based on “soft” management practices.  2000
Elsevier Science Ltd. All rights reserved.

Keywords: Continuous Improvements; Innovation; Manufacturing; Australia

1. Introduction implement on-going product, service, and process inno-

1.1. Background
Organisations can build competitive advantage
through superior manufacturing or service delivery, but
sustaining the competitive advantage over time requires
comparable skills in developing a continual stream of
new products and services. The increasing pace of tech-
nological change and the accelerating globalisation of
business has meant that competitive advantage for many
corporations now lies in their ability to effectively
* Corresponding author.
E-mail address: mile.terziovski@buseco.monash.edu.au (M.
Terziovski).
1
Present address: Euro-Australian Cooperation Centre—Mel-
bourne and Senior Lecturer, Department of Management, Faculty of
Economics and Commerce, University of Melbourne, Royal Parade,
Parkville, Victoria 3052, Australia.
vations. As product innovation cycles become shorter
and more frequent, and innovation becomes a dominant
strategic weapon, companies will be forced to exploit
synergies between products, services and processes. As
product innovation is a knowledge-based process, this
requires mastering the overall process of knowledge cre-
ation, dissemination and application. This progressive
accumulation and sharing of knowledge fosters the pro-
cess of organisational learning that is the essential engine
for the continuous improvement process. Hence, long
term competitiveness is increasingly dependent on how
well a company can continuously improve its product
development capabilities by fostering organisational
learning and utilising individual and group knowledge
within the company.
1.2. The continuous improvement concept
The continuous improvement concept is driven by the
Deming Cycle (Evans and Lindsay, 1999) and the

0166-4972/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 6 6 - 4 9 7 2 ( 9 9 ) 0 0 1 7 3 - X
540 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550
Kaizen concept (Imai, 1986). This is a methodology for
continuous improvement, composed of four stages: Plan,
Do, Check, and Act. The Plan stage consists of studying
the current situation, gathering data and planning for
improvement. In the Do stage, the plan is implemented
on a trial basis. The Check stage is designed to determine
if the trial plan is working correctly and if any further
problems or opportunities are found (Imai, 1986). The
last stage, Act, is the implementation of the final plan
to ensure that the improvements will be standardised and
practiced continuously. This leads back to the Plan stage
for further diagnosis and improvement.
Imai’s best selling book, Kaizen: The Key to Japan’s
Competitive Success (Imai, 1986) shows how the orig-
inal Western concepts have been adapted to the Japanese
culture to provide the key to Japan’s post war success.
As a direct consequence of the Total Quality Control
(TQC) philosophy (Feigenbaum, 1983), Kaizen
(continuous improvement) strategy has a large-scale par-
ticipatory dimension by all employees in an organisation.
This participatory dimension is not entirely new (Merli,
1990). Western authors such as Likert (1967) had
already formulated participatory management before it
developed in Japan. However, Likert’s participatory
management theory is one example. Kaizen is more
strongly oriented towards continuous improvement than
towards management. Imai (1986) broadly described the
Kaizen strategy to include concepts, systems, and tools
within the bigger picture of leadership involvement and
people culture — all driven by the customer. Imai (1986,
p. 3) defined Kaizen as follows:
The essence of Kaizen is simple and straightfor-
ward: Kaizen means improvement. Moreover, Kaizen
means ongoing improvement involving everyone,
including both managers and workers.
The outcome of the Kaizen Strategy are improvements
in Quality, Cost, and Delivery. The underlying principle
of the Kaizen strategy is the recognition that manage-
ment must seek to satisfy the customer and serve cus-
tomer needs if it is to stay in business and make a profit.
Improvements in such areas as quality, cost, and sched-
uling (meeting volume and delivery requirements) are
essential. Kaizen is a customer-driven strategy for con-
tinuous improvement. Therefore, it is assumed that all
Kaizen activities should eventually lead to increased cus-
tomer satisfaction.
1.3. Implementing Kaizen
The underpinning principle of Kaizen is the use of
various problem-solving tools for the identification and
solution of work-based problems. The aim is for
improvement to reach new “benchmarks” with every
problem that is solved. To consolidate the new bench-
mark, the improvement must be standardised. In many
Australian firms this standardisation has been attempted
via the ISO 9000 quality systems certification. Kaizen
generates process-oriented thinking (P criteria)since pro-
cesses must be improved before improved results (R
criteria) can be obtained. Improvement can be broken
down between continuous improvement and innovation.
Kaizen signifies small improvements made in the status
quo as a result of ongoing efforts. On the other hand
innovation involves a step-change improvement in the
status quo as a result of a large investment in new tech-
nology and/or equipment.
There is one significant difference between Kaizen
and Innovation. Kaizen does not necessarily call for a
large investment in capital to implement the strategy.
However, the Kaizen strategy does call for continuous
effort and commitment from all levels of management.
Thus Kaizen calls for a substantial management commit-
ment of time and effort. Investing in Kaizen means
investing in people. According to Imai (1986, p. 217)
the Kaizen initiatives that have been implemented in
Japan have had one key practice in common. That is,
overcoming employees’ resistance to change. This was
achieved by addressing the following critical issues:
1. Constant effort to improve industrial relations.
2. Emphasis on training and education of employees.
3. Developing informal leaders among the workers.
4. Formation of small-group activities such as QC
circles and improvement teams.
5. Support and recognition for workers’ Kaizen efforts
(P criteria).
6. Efforts for making the workplace a place where
employees can pursue goals.
7. Bringing social life into the workplace as much as
practical.
8. Training supervisors so that they can communicate
better with workers and can create a more positive
involvement with workers.
According to Imai (1986, p. 204). “Unless top manage-
ment is determined to introduce Kaizen as a top priority,
any effort to introduce Kaizen into the company will be
short-lived.”
This paper investigates the adoption of Continuous
Improvement (CI) strategies and their impact in Aus-
tralian manufacturing firms. This is the first empirical
study conducted in Australia that has focussed on CI.
Data was collected by means of a postal questionnaire
survey that was mailed to 1200 manufacturing firms.
This survey is part of a global study that has examined
CI strategies in over ten countries. The Australian survey
resulted in 385 responses. Our analysis focuses on the
following six aspects of CI in Australian manufactur-
ing firms:
 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550
1. The development of a set of measures to gauge the
effect of CI.
2. Determining the success of CI by identified the struc-
tural variables pertaining to the organisation.
3. Examining how the integration of CI influences the
performance of the organisation.
4. Identifying the motivations for implementing a CI
program.
5. Examining the impact of various approaches to CI.
6. Identifying the tools utilised in the CI process and the
relationship between these tools and organisational
performance.
The remainder of this paper is structured as follows. The
next section briefly describes the survey methodology.
The results of our analysis, along the dimensions dis-
cussed above, are then presented. The paper concludes
with a discussion of the findings and implications for
management.
2. Methodology
The questionnaire used for data gathering in Australia
was based on an instrument used for similar studies con-
ducted in ten European countries. The design of the
questionnaire was finalised using a pilot survey to test
the validity of the questionnaire for the Australian manu-
facturing industry. For the mail survey, a sample of 1200
manufacturing companies was selected. The survey
scope was limited to manufacturing companies with
greater than A$10 million annual sales turnover. A dif-
ferent research group in each state in Australia managed
the administration of the questionnaire. The response
rate varied for each state, ranging from 25% to 49%.
Overall, 385 firms responded, giving a response rate of
32%. Responding firms were classified by state and
industry sectors.
3. Analysis
3.1. Measuring the success of CI processes
Two scales were developed to gauge the success of
CI processes. The first scale related to the overall per-
formance of each organisation. Specifically, respondents
rated the extent to which productivity, manufacturing
quality, delivery performance, lead-time, and product
cost had changed during the past two years. The average
of these five ratings was utilised to measure organis-
ational performance. Cronbach’s alpha was 0.80
(n=184), reflecting an encouraging level of internal con-
sistency.1 The second scale corresponded to previous
1
Wilks (1946) test of psychometric parallelism was also conducted.
According to this procedure, the scale was not unidimensional, χ2
(17)=205.314. Nonetheless, all of the items clearly relate to organis-
541
experiences with CI. Respondents rated past experiences
with CI efforts on four aspects: initiating concrete
changes, maintaining activities in on-going efforts,
spreading change efforts to other departments and units,
and managing several projects simultaneously. Each
aspect was rated on a five-point scale, where 1 rep-
resented very negative experiences and 5 represented
very positive experiences. The average of these four
items was used to measure past experience. Cronbach’s
alpha was 0.76 (n=385), demonstrating an adequate level
of internal consistency.
3.2. The influence of organisational structure on CI
success
The extent to which the measures of CI success
(organisational performance and past experiences) are
influenced by organisational structure were examined
using the following five structural variables:
1. Annual turnover of the business unit
2. Number of production workers in the business unit
3. Number of non-production workers in the business
unit
4. Percentage of products that are entirely unique, that
is, designed and manufactured to customer order.
5. Percentage of products that are modulised, with mod-
erate customisation to order.
Table 1 displays the Pearson product-moment corre-
lation between each of these structural variables and the
two measure of CI success. This table reveals that organ-
isational performance positively correlated with the per-
centage of unique products. In other words, organis-
ations that frequently design and manufacture products
to satisfy specific orders tended to perform more effec-
tively. However, none of the other correlations departed
significantly from zero. That is, the impact of CI was
independent of annual sales or number of employees.
3.3. Extent of integration of CI
This section examines the extent to which CI has been
integrated within the organisation. That is, the degree to
which employees and management were involved in the
CI process. The extent to which integration influences
the impact of CI is also discussed. Eight aspects of inte-
gration were utilised. Table 2 provides the mean and
standard deviation for seven of these aspects. This table
provides some illuminating results. First, the average
level of maturity approximated 3.5. According to the
scale provided to respondents, this level corresponds to
ational performance. Hence, the presence of multidimensionality does
not undermine the interpretations.
542 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550

Table 1
Bivariate correlation between the structural variables and the measures of CI success

Organisational performance Past experiences

Annual turnover ⫺0.01 0.04

Number of production workers ⫺0.09 0.05
Number of non-production workers ⫺0.03 0.03
Percentage of unique products 0.17a ⫺0.06
Percentage of modularised products ⫺0.04 0.01

a
P⬍0.05.

Table 2
Mean and standard deviation for seven aspects pertaining to the extent of integration

Mean Standard deviation

Length of time CI has been utilised (years) 5.04 8.82

Level of maturity (out of 10) 3.53 1.86
Percentage of operators actively involved in CI 31.57 28.62
Percentage of non-operators actively involved in CI 33.61 27.64
Percentage of employees trained in problem solving 27.11 26.82
Number of ideas registered in the past two years 131 343
Number of ideas implemented in the past two years 73 240

the learning stage. Second, the percentage of operators
and non-operators actively involved in CI also provided
some invaluable information. To investigate this issue,
both of these variables were subjected to an arcsine
transformation (Cohen, 1983). A related t–test then com-
pared these transformed variables. This procedure
revealed that non-operators are more likely to be
involved in the CI process, t(239)=2.15, P⬍0.05, albeit
to a trivial extent. Finally, only about half of the regis-
tered ideas and suggestions had been implemented by
the respondents’ organisations.
The final aspect of integration concerned the breadth
of CI. This aspect distinguishes between those organis-
ations in which CI is used in all areas and those organis-
ations in which CI is used in a restricted number of areas.
Fig. 1 displays a frequency distribution associated with
the responses to this issue. Almost 50% of the organis-
ations have applied CI to the entire business.
Canonical correlation analysis was implemented to
ascertain the relationship between the two measures of
CI success and the various aspects of integration. Those
aspects corresponding to percentages were subjected to
an arcsine transformation. Breadth of CI was measured
on a 5-point scale, where 1 represented organisations
that have not employed CI and 5 represented organis-
ations that have employed CI throughout the entire busi-
ness. Finally, to circumvent potential multicollinearity,
number of ideas for improvements registered was
excluded from the analysis. The first canonical corre-
lation between CI success and integration approximated
0.6. According to Bartlett’s (1941) test, this correlation
significantly exceeded zero, χ2(14)=31.03, P⬍0.01. The
remaining canonical correlation, however, did not attain
significance, χ2(6)=10.90, P⬎0.05. Table 3 provides the
correlation between each variable and the corresponding
canonical variate. Essentially, these correlations reflected
the contribution of each variable to the canonical corre-
lation. In other words, these correlations represented the
importance of each variable. These results indicated that
past experiences with CI vary with every aspect of inte-
gration, apart from the number of ideas implemented.
Specifically, positive experiences with CI tend to
accompany mature and broad CI programs, a higher per-
centage of employees actively involved in the program,
and a greater emphasis on training in problem solving.
In summary, previous experiences are positively related
to the degree of implementation.

Fig. 1. Frequency of the breadth of CI integration.

 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550 543

Table 3
Correlation between each variable and the corresponding canonical variate

Correlation with corre- Correlation with correspond-

Measure of CI success Aspect of integration
sponding variatea ing variatea

Organisational performance 0.030 Length of time CI has been utilised 0.743*

Past experiences 0.966* Level of maturity 0.669*
Percentage of operators involved 0.527*
Percentage of non-operators involved 0.560*
Percentage of employees trained in problem
0.705*
solving
Number of ideas implemented 0.285
Breadth of CI 0.467*

a
*r⬎0.45

3.4. Motivation for implementing a CI program
This section explores the motives underlying the
adoption of CI. Respondents were presented with a list
of 13 possible motives for CI. Each motive was rated
on a 5-point scale, where 1 represented no importance
and 5 represented critical importance. The mean and
standard deviation pertaining to each motive is presented
in Table 4. The most pervasive motivations related to
improved quality conformance, increased manufacturing
productivity, reduced costs, and improved delivery
reliability.
Factor analyses were conducted to identify those motiv-
ations that tend to co-occur. This process entailed two
phases. First, a principal components analysis, with vari-
max rotation, was applied to a random selection of 201
respondents. This phase was implemented to uncover
sets of related items. Second, a confirmatory factor
analysis was applied to the remaining 201 respondents,
using EQS for Windows program (see Bentler and Wu,
1995). This phase was implemented to corroborate the
findings derived from the principal components analysis.
The principal components analysis extracted two inter-
pretable factors. Taken together, these factors explained
almost 40% of the pre-rotation variance. Table 5 pro-
vides the outcome generated by the factor analyses. The
first column identifies each motivation. The second col-
umn provides the factor loadings pertaining to the first
component or factor. The second column provides the
factor loadings pertaining to the second component. The
remaining columns correspond to the confirmatory factor
analysis and will be explicated later. Only those factor
loadings that exceed 0.4 are displayed. Motivations that
load on neither component have been excluded from
this table.
The first component related to augmenting production
volume and manufacturing productivity, and to attenuat-
ing production lead times and cost. This component was
referred to as “Production efficiency”. The second
component related to increasing employee commitment
and skills, and to enhancing organisation, cooperation,
and communication. This component was designated as
“Employee performance”. The confirmatory factor
analysis demonstrated that both of these factors general-

Table 4
Mean and standard deviation associated with each motive. Higher numbers represent greater importance

Mean Standard deviation No. of responses

Because our customers ask for CI 2.33 1.41 172

Increase production volume 3.23 1.20 174
Increase manufacturing productivity 4.06 0.94 177
Improve quality conformance 4.19 0.99 183
Reduce production lead times 3.40 1.22 181
Improve delivery reliability 3.83 1.11 179
Improve safety and physical environment 3.35 1.21 179
Cost reduction 3.97 1.09 186
Improve administration routines 2.67 1.17 175
Increase employee commitment/attitude towards
3.61 1.10 179
change
Improve organisation, cooperation and communi-
3.70 1.05 181
cation
Increase employee skills 3.30 0.99 176
Because CI is a management directive 1.96 1.20 168
544 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550

Table 5
Results derived from the principal components analysis and the confirmatory factor analysis applied to motives for CI

Principal components analysis Confirmatory factor analysis

First component Second component First factor Second factor

Increase production volume 0.41 0.60

Increase manufacturing productivity 0.77 0.84
Reduce production lead times 0.42 0.49
Cost reduction 0.81 0.50
Increase employee commitment/ attitude
0.86 0.81
towards change
Improve organisation, cooperation and com-
0.76 0.82
munication
Increase employee skills 0.75 0.70
Percentage of pre-rotation variance explained 26.8 11.6
Pre-rotation eigenvalues 3.468 1.513

ised to the remainder of this sample. This analysis is
summarised in Table 6. In particular, this table provides
four goodness-of-fit indices pertaining to several factor
structures. Specifically, the four indices include the χ2
statistic, the Bentler–Bonnet normed fit index (NFI), the
Bentler–Bonnet nonnormed fit index (NNFI), and the
mean value of the absolute standardised off-diagonal.
Low χ2 values and standardised residuals, together with
high NFI and NNFI values, reflect a superior factor
structure. In particular, NFI and NNFI values should ide-
ally exceed 0.9.
Four factor structures or models were investigated.
The null model presupposes that all motivations are
independent. In contrast, the one-factor model assumes
that each motivation corresponds to the same factor. The
two-factor orthogonal model assumes that motivations
relate to one of two independent factors. The output
derived from the principal components analysis was used
to identify the motivations that pertain to each factor.
Finally, the two-factor oblique model presupposes that
all motivations correspond to one of two correlated fac-
tors. According to the goodness-of-fit indices displayed
in Table 6, the two-factor oblique model was clearly bet-
ter than each of the remaining structures. That is, this
model corresponds to the lowest χ2 value and mean
residual, and the highest NFI and NNFI values. The third
and fourth columns in Table 5 present the standardised
coefficients pertaining to the two-factor oblique model.
All of these coefficients exceed 0.48, which further cor-
roborates the efficacy of this structure. Finally, the corre-
lation between the two factors approximated 0.31.
In short, the factor analyses revealed two correlated
factors: Production efficiency and Employee perform-
ance. Cronbach’s alpha associated with production
efficiency was 0.67 (n=328), which reflects a reasonable
level of internal consistency given the small number of
items pertaining to this factor. Cronbach’s alpha per-
taining to Employee performance was 0.82 (n=342),
which is clearly encouraging. These findings reveal that
issues associated with production efficiency primarily
motivated some organisations, whereas issues associated
with employee performance primarily motivated other
organisations. To ascertain whether or not motivations
for the adoption of CI influence its success, a canonical
correlation analysis was undertaken. Two motivation
scores were calculated for each organisation. The first
score corresponded to the degree to which production
efficiency motivated the CI program. The second score
corresponded to the degree to which employee perform-
ance motivated the CI program. Both of these scores
were generated by averaging the responses on those
items relating to each factor. The canonical correlation
between the motivation factors and CI success, however,
did not attain significance, r(312)=0.238, χ2(4)=8.829,
P⬎0.05. This finding suggests that an organisation’s
motivation for adopting CI does not influence perform-
ance or success.

Table 6
Goodness-of-fit indices pertaining to four factor structures

Model Degrees of freedom χ2 NFI NNFI Mean residual

Null 21 309.761 – – –
One-factor 14 106.277 0.657 0.521 0.096
Two-factor orthogonal 14 43.956 0.858 0.844 0.099
Two-factor oblique 13 24.956 0.919 0.933 0.057
 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550 545

Table 7
Mean and standard deviation associated with each issue. Higher numbers represent issues that are addressed frequently in the CI process

Mean Standard deviation No. of responses

Machine up-time speed 2.68 1.39 324

Set-up time 2.67 1.37 326
Work methods and tools 3.33 1.17 345
Manufacturing quality 3.90 1.11 344
Material waste 3.37 1.29 341
Product design 2.86 1.33 337
Administrative routines 2.66 1.19 345
Safety 3.21 1.31 332
Manufacturing costs 3.87 1.16 341
Buffer stock, lay-out and physical flows 2.87 1.27 334
Employee commitment 3.30 1.22 342
Employee skills 3.31 1.12 343

3.5. Approaches to CI: content and responsibility
This section identifies current approaches to the CI
process and whether or not CI success depends on the
approach that is adopted. First, respondents were
presented with a list of 12 issues that could potentially
be addressed in the CI process. Each issue was rated on
a five-point scale, where 1 represented issues that were
not discussed frequently in the CI process and 5 rep-
resented issues that were discussed very frequently. The
mean and standard deviation pertaining to each issue is
displayed in Table 7. The issues discussed most pervas-
ively included manufacturing quality and costs. Admin-
istrative routines and set-up times were addressed the
least often.
Factor analyses were conducted to identify sets of
related issues. Again, a principal components analysis
was applied to a random sample of 201 respondents. A
confirmatory factor analysis was subsequently applied to
the remaining participants. Table 8 provides the outcome
of these analyses.
The principal components analysis uncovered two
interpretable factors or components. These components
explained almost 45% of the pre-rotation variance. The
first component related to machine up-time speed, set-
up time, and manufacturing costs. This component was
referred to as “Efficiency”. The second component
related to safety, employee commitment, and employee
skills. This component was designated as “Employee
environment”. These results seem to indicate that some
organisations primarily address issues related to
efficiency, whereas other organisations tend to address
issues related to the employees. Table 9 provides good-
ness-of-fit indices associated with four factor structures
or models. Again, the four structures included the inde-
pendence model, a one-factor model, an orthogonal two-
factor model, and an oblique two-factor model.
According to the goodness-of-fit indices displayed in
Table 9, the two-factor oblique model is patently
superior to each of the remaining structures. That is, this
model corresponds to the lowest χ2 value and mean
residual, and the highest NFI and NNFI values. In
addition, the NFI and NNFI approach 0.9, thereby
reflecting an adequate model. The third and fourth col-
umns in Table 7 report the standardised coefficients per-
taining to the two-factor oblique model. All of these
coefficients exceed 0.5, which further substantiates the
appropriateness of this factor structure. Finally, the cor-
relation between the two factors approximated 0.32. To
reiterate, the factor analyses revealed two correlated fac-

Table 8
Results derived from the principal components analysis and the confirmatory factor analysis applied to the issues addressed in the CI process

Principal components analysis Confirmatory factor analysis

First component Second component First factor Second factor

Machine up-time speed 0.84 0.79

Set-up time 0.83 0.69
Safety 0.60 0.51
Manufacturing costs 0.51 0.50
Employee commitment 0.69 0.83
Employee skills 0.74 0.85
Percentage of pre-rotation variance
30.7 13.4
explained
Pre-rotation eigenvalues 3.688 1.606
546 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550

Table 9
Goodness-of-fit indices associated with four factor structures

Model Degrees of freedom χ2 NFI NNFI Mean residual

Null 15 253.742 – – –
One-factor 9 71.783 0.717 0.562 0.081
Two-factor orthogonal 9 36.180 0.857 0.810 0.126
Two-factor oblique 8 24.325 0.904 0.872 0.064

tors: Efficiency and Employee environment. Cronbach’s
alpha pertaining to efficiency was 0.70 (n=319), which
reflected an acceptable level of internal consistency.
Cronbach’s alpha pertaining to Employee environment
was 0.72 (n=326), which also indicated a reasonable
level of internal consistency. Apart from the issues
addressed by CI, the survey also investigated employees
responsible for the process. Specifically, responsibility
could be allocated to business unit managers, manage-
ment teams, manufacturing managers, and so on. Table
10 provides the percentage of organisations that allotted
responsibility to each individual group. In general, the
business unit manager or management team set the direc-
tion. Management teams, department managers, and
supervisors were primarily responsible for monitoring
and co-ordinating the process. Finally, supervisors and
operators tended to implement CI.
To ascertain whether or not the allocation of responsi-
bility influences the success of CI, two phases were con-
ducted. The first phase was implemented to select a sub-
set of individuals or groups that could determine the
success of CI. Towards this end, 60 independent t-tests
were conducted. Each t-test determined whether or not
assigning responsibility to a particular individual or
group influenced organisational performance or past
experiences with CI. For instance, one t-test compared
performance on those organisations in which business
unit managers set the direction with performance on the
remaining organisations. In seven cases, assigning
responsibility to a particular individual or group signifi-
cantly enhanced success at the 0.01 level. These cases
included (a) management teams setting direction, (b)
improvement teams setting direction, (c) supervisors
monitoring the process, (d) operators monitoring the pro-
cess, (e) supervisors implementing CI, (f) operators
implementing CI, and (g) regular work teams
implementing CI.
During the second phase, a canonical correlation
analysis was implemented. This analysis determined the
relationship between CI success and the approach
adopted. Organisational performance and past experi-
ences were employed to gauge success. The seven cases
discussed in the previous paragraph, together with the
two factors associated with the issues addressed, were
used to define the approach. The first canonical corre-
lation between CI success and the approach adopted
approximated 0.42. According to Bartlett’s (1941) test,
this correlation significantly exceeded zero,
χ2(18)=42.439, P⬍0.01. The remaining canonical corre-
lation, however, did not attain significance,
χ2(8)=12.549, P⬎0.05. Table 11 presents the correlation
between each variable and the corresponding canonical
variate. Recall that all of these correlations effectively
represent the importance of each variable. These results
indicate that both organisational performance and past
experiences depend on the approach adopted. Specifi-
cally, CI was more successful when issues relating to
efficiency and employees were addressed frequently,
improvement teams were responsible for setting direc-
tion, supervisors or operators monitored the process, and
regular work teams implemented CI.

Table 10
Percentage of organisations who assigned responsibility to the various groups

Who sets direction? Who monitors and coordinates Who carries out CI

Business unit manager 50.0 18.6 9.8

Management team 53.1 38.8 20.2
Manufacturing manager 26.2 31.9 20.0
Staff function manager 8.6 18.3 22.9
Department managers 19.5 38.8 35.0
Supervisors 10.5 27.4 51.9
Operators 5.2 11.4 51.0
Regular work teams 5.7 13.3 37.1
CI facilitator/coordinator 11.7 28.1 16.4
Improvement teams 10.5 18.6 36.7
 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550 547

Table 11
Correlation between each variable and the corresponding canonical variate

Correlation with corre- Correlation with corresponding

Measure of CI success Approach adopted
sponding variatea variatea

Organisational performance 0.981* Employee environment 0.702*

Past experiences 0.405* Efficiency 0.537*
Management teams setting direction 0.376
Improvement teams setting direction 0.427*
Supervisors monitoring 0.553*
Operators monitoring 0.402*
Supervisors implementing CI 0.074
Operators implementing CI 0.248
Regular work teams implementing CI 0.566*

a
*r⬎0.4

3.6. Tools utilised in the CI process
This final research section identified the tools that are
currently employed to undertake CI. In addition, the
impact of these tools on CI success was also explored.
These tools were subdivided into three classes: general
tools, such as means of promoting changes and support,
problem-solving tools, and incentive tools.
First, respondents were presented with a list of 16 gen-
eral tools. Each tool was rated on a five-point scale,
where 1 represented not important to the CI process and
5 represented critical importance. Table 12 reports the
mean and standard deviation for each tool. Supportive
leadership, support for managerial staff, monitoring the
CI-process and face-to-face communication were found
to be the most important tools.
Factor analyses were undertaken to identify sets of
related tools. Again, a principal components analysis
was applied to a random sample of 201 participants. A
confirmatory factor analysis was subsequently applied to
the remaining respondents. Table 13 displays the out-
come of these analyses. The principal components analy-
sis uncovered three interpretable factors. These compo-
nents explained over 52% of the pre-rotation variance.
The first components relate to working in teams, sugges-
tion schemes, and promotions via notice boards and
internal media. This factor will hereafter be called
“Group communication”. The second component relates
to training in problem-solving, monitoring, support for
management, and supportive leadership. This factor will
be referred to as “Coordinating CI”. The final component
pertains to use of ISO 9000, Total Productive Mainte-
nance, and formal policy deployment and will be called
“Formal systems”.
Table 14 displays goodness-of-fit indices pertaining
to five factor structures or models. The five structures
included the independence model, an orthoginal two-fac-
tor orthogonal model, an oblique two-factor model, and
an orthogonal three-factor model and an oblique three
factor model. The two factor models were generated by

Table 12
Mean and standard deviation associated with each tool. Higher numbers represent tools that are deemed as more important

Mean Standard deviation No. of responses

Training of personnel in problem solving skills 3.69 1.17 341

Monitoring the CI-process 4.18 2.26 341
Support for managerial staff 4.30 0.98 345
Incentive systems 2.35 1.21 336
Supportive leadership 4.37 3.05 343
Work in teams/work groups 3.74 1.16 342
A suggestion scheme 2.37 1.17 330
A general problem solving format 2.88 1.31 325
Promoting on notice boards 2.57 1.19 333
Promoting through internal media 2.51 1.17 327
Promoting through competitions and awards 2.09 1.13 323
Face to face communication 4.03 .96 342
Regular shop floor visits by management 3.81 1.15 333
Use of ISO 9000 3.51 1.35 337
Use of Total Productive Maintenance 2.70 1.24 316
Use of formal policy deployment 2.81 1.23 325
548 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550

Table 13
Results derived from the principal components analysis and the confirmatory factor analysis applied to the tools utilised in the CI process

Principal components analysis Confirmatory factor analysis

First Second Third First Second Third

Training of personnel in prob-

0.73 0.78
lem solving skills
Monitoring the CI-process 0.71 0.69
Support for managerial staff 0.68 0.70
Supportive leadership 0.66 0.62
Work in teams/work groups 0.53 0.59
A suggestion scheme 0.64 0.58
Promoting on notice boards 0.77 0.82
Promoting through internal
0.84 0.85
media
Use of ISO 9000 0.78 0.83
Use of Total Productive Main-
0.50 0.51
tenance
Use of formal policy deploy-
0.62 0.54
ment
Percentage of pre-rotation
27.7 15.5 8.9
variance explained
Pre-rotation eigenvalues 4.43 2.47 1.43

Table 14
Goodness-of-fit indices associated with five factor structures

Model Degrees of freedom χ2 NFI NNFI Mean residual

Null 55 381.513 – – –
Two-factor orthogonal 44 97.014 0.746 0.682 0.128
Two-factor oblique 43 95.601 0.749 0.679 0.091
Three-factor orthogonal 44 64.211 0.832 0.790 0.138
Three-factor oblique 41 31.542 0.917 0.889 0.054

rotating only two components in the principal compo-
nents analysis.
According to the goodness-of-fit indices displayed in
Table 14, the three-factor oblique model is better than
each of the remaining factor structures. This model cor-
responds to the lowest χ2 value and mean residual, and
the highest NFI and NNFI values. Moreover, the NFI
and NNFI approach 0.9, thereby reflecting an adequate
model. The third and fourth columns in Table 13 provide
the standardised coefficients pertaining to the three-fac-
tor oblique model. All of these coefficients exceed 0.5,
which further substantiates the appropriateness of this
factor structure. The correlation between Group com-
munication and Co-ordinating CI was 0.38. In contrast,
the correlation between Co-ordinating CI and Formal
systems was only 0.19 and the correlation between
Group communication and Formal systems was 0.21.
Taken together, these findings suggest that some organ-
isations primarily utilise formal systems, whereas other
organisations are more concerned with communication
or co-ordination. In short, the factor analyses generated
three correlated factors. Cronbach’s alpha associated
with Group communication was 0.66 (n=321), which
reflects a moderate level of internal consistency. Cron-
bach’s alpha relating to Co-ordinating CI was more
encouraging and approximated 0.78 (n=329). Finally,
Cronbach’s alpha associated with Formal systems was
only 0.62 (n=30 726). The prevalence of tools relating
to problem-solving and incentives was also investigated.
Table 15 provides the percentage of organisations that
employed each tool often, occasionally, or never. The
problem-solving tools that are used most predominantly
relate to identification, the seven basic quality tools, and
process-mapping. CI tends to be rewarded through career
development and, to a lesser extent, through individual
salaries.
A series of one-way ANOVAs were conducted to sel-
ect a subset of problem-solving tools and incentives that
could potentially enhance CI success. Seven tools and
incentives were found to significantly improve CI per-
formance at the 0.001 level: the seven new quality tools,
FMEA, QFD, creativity tools, standardisation tools, 5S,
and CI rewarded through career development. Nonethe-
less, some of these relationships could be spurious.
To circumvent this potential shortfall, another canoni-
cal correlation analysis was conducted. This analysis
 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550 549

Table 15
Percentage of organisations that utilise each problem-solving tool or incentive

Usage
Often Occasionally Never

Problem solving tools

Problem identification tools/checklists 61.8 28.9 9.3
Seven basic quality tools 36.1 35.4 28.5
Seven new quality tools 8.4 17.1 74.5
Process mapping tools 31.3 34.0 34.7
Failure mode and effect analysis (FMEA) 16.4 21.3 62.2
Quality function deployment (QFD) 14.1 21.0 64.9
Creativity tools/idea generation tools 23.0 35.3 41.7
Display/visualisation tools 26.8 31.7 41.5
Standardisation tools 26.6 28.1 45.3
5S 16.9 16.9 66.3
Incentives
Suggestions are evaluated and rewarded 24.6 35.3 40.1
CI activities are rewarded directly through one
12.0 16.8 70.9
off bonuses
CI activities are rewarded indirectly through
30.3 25.0 44.7
individual salaries
CI is not rewarded monetarily, but through
40.5 38.3 21.2
development of individual jobs, careers

investigated the relationship between CI success and the
tools employed. Organisational performance and pre-
vious experiences were again utilised to gauge CI suc-
cess. The tools examined included Group communi-
cation, Coordinating CI, Formal systems, and the seven
problem solving tools and incentives specified in the pre-
vious paragraph. The first canonical correlation between
CI success and the tools utilised approached 0.55.
According to Bartlett’s (1941) test, this correlation
significantly departed from zero, χ2(20)=46.192,
P⬍0.01. The remaining canonical correlation, however,
did not attain significance, χ2(9)=6.054, P⬎0.05. Table
16 presents the correlation between each variable and
the corresponding canonical variate. These results indi-
cate that both organisation performance and past experi-
ences related to the problem-solving tools that were
adopted. Specifically, the seven new quality tools,
FMEA, QFD, creativity tools, standardisation tools, and
5S all enhanced CI success.
4. Discussion and implications for managers
The analysis shows that respondent organisations pre-
ferred the use of techniques such as problem identifi-
cation tools, checklists and the seven basic quality tools
rather than the more advanced problem solving tech-
niques such as failure mode and effect analysis (FMEA),
quality function deployment (QFD) or so-called “soft”
options which relate to people and teams. Other Aus-
tralia research by Terziovski et al. (1999) indicates that
areas such as customer surveys, leadership and com-

Table 16
Correlation between each variable and the corresponding canonical variate

Correlation with correspond- Correlation with corresponding

Measure of CI success Approach adopted
ing variatea variatea

Organisational performance 0.903* Group communication 0.226

Past experiences 0.611* Coordination of CI ⫺0.12
Formal systems 0.147
Seven new quality tools 0.500*
FMEA 0.588*
QFD 0.703*
Creativity tools 0.771*
Standardisation tools 0.653*
5S 0.653*
CI rewarded through career development 0.486

a
*r⬎0.5.
550 M. Terziovski, A.S. Sohal / Technovation 20 (2000) 539–550
munication, and other qualitative interventions tend to
be used in preference to quantitative quality management
tools requiring numerical skills. Motivational factors for
introducing CI included production efficiency and
improved quality of performance. The implications of
the current survey findings suggest that the majority of
the organisations participating in the study have intro-
duced CI to at least part of their operations. For approxi-
mately half of the respondents, CI appears to be quaran-
tined to manufacturing/production rather than being
applied organisation wide. Organisational performance
was found to be linked to the length of time CI had been
implemented and the extent of CI involvement. Organis-
ations with partial implementation of CI will need to
examine more extensive development of CI through
other organisational systems in addition to those areas
devoted to manufacturing and production. These findings
have implications for the large, multinational corpora-
tions as well as the small–medium enterprises (SMEs).
In the first instance there is a need for large corporations
to develop strategic plans which support product devel-
opment and innovation, communication networks, and
disciplined, sophisticated problem solving across all
business units. This includes international as well as
national sites involved in the development and pro-
duction of a particular product range, as well as different
companies owned by the corporation. Furthermore busi-
ness units need to introduce CI across all functional
areas, not just manufacturing and production. This will
also further enhance cross-national learning to facilitate
Innovation Management, by enabling the organisation to
develop a global pool of skills and resources. Globalis-
ation for SMEs may include joint venture projects, with
individual organisations being responsible for a parti-
cular aspect of the project. These enterprises need to not
only have effective CI processes within their own organ-
isation, but also establish links between the product and
process development, communication networks, and
problem solving strategies of the partners in the venture.
References
Bartlett, M.S., 1941. The statistical significance of canonical corre-
lations. Biometrika 32, 29–38.
Bentler, P.M., Wu, E.J.C., 1995. EQS for Windows User’s Guide.
Multivariate Software, Inc, Encino (CA).
Cohen, J., 1983. Applied Multiple Regression/Correlation Analysis for
the Behavioural Sciences. Erlbaum, Hillsdale (NJ).
Evans, J.R., Lindsay, W.M., 1999. The Management and Control of
Quality, 4th ed. West Publishing Company.
Feigenbaum, A.V., 1983. Total Quality Control, 3rd ed. McGraw-Hill,
New York.
Imai, M., 1986. Kaizen: The Key to Japan’s Competitive Success.
McGraw–Hill, New York.
Likert, R., 1967. The Human Organisation. McGraw–Hill, New York.
Merli, G., 1990. Total Manufacturing Management: Production Organ-
isation for the 1990s. Productivity Press, Cambridge.
Terziovski, M., Sohal, A.S., Moss, S., 1999. Longitudinal analysis of
quality management practices in Australian organisation. Total
Quality Management 10 (6), 915–926.
Wilks, S.S., 1946. Sample criteria for testing the equality of means,
equality of variances, and equality of covariances in a normal
multivariate distribution. The Annals of Mathematical Statistics 17,
257–281.
Milé Terziovski is a multi-disciplined pro-
fessional with 16 years experience with RTZ.
He has obtained a diversified set of academic
qualifications from four Australian Universities:
PhD, MBA, M.Eng(Hons), B.Eng(Hons),
Grad.Dip.Mgmt. Dr Terziovski holds a Senior
Lecturer (tenured) post with the Department of
Management, Faculty of Business and Econom-
ics, Monash University where he teaches Stra-
tegic Management and Quality Management
subjects and Re-engineering the Organisation,
for the Monash Mt Eliza Business School, MBA
Program. Dr Terziovski is the Foundation Director of the Euro–Australian
Cooperation Centre (EACC) for Continuous Improvement and Innovation
Management, Victoria, and Deputy–Chair of the Board of Management for
the EACC–Australia. Dr Terziovski is an active researcher in innovation
management, strategy implementation, continuous improvement, quality
management, ISO 9000 certification, supply chain management, and
outsourcing/business process re-engineering in the manufacturing and ser-
vices sectors for both large and small to medium Enterprises (SME’s). He
has received large research grants from the State (Strategic Industry
Research Foundation) and Federal Government (Department of Industry,
Science and Resources), the Australian Research Council and Monash
University. He has published over 60 articles, book chapters and working
papers. He has commercially published his PhD thesis and has edited a
collection of book chapters based on the Second International and Fifth
National Research Conference on Quality Management, which he chaired
during 1998.
Amrik S. Sohal is a Professor in the Depart-
ment of Management, Faculty of Business and
Economics at Monash University, Melbourne,
Australia, where he is also Director of the Qual-
ity Management Research Unit. From mid-1993
to mid-1997, Professor Sohal served as an
Associate Dean (Research Development) and
Associate Dean (Graduate Teaching) for the
Faculty of Business and Economics at Monash
University. He holds a PhD in Manufacturing
Operations Management from the University of
Bradford Management Centre in the UK, as well
as a BEng (Hons) and an MBA, also from the University of Bradford. He
is a Chartered Engineer and a member of the Institution of Electrical
Engineers.
Professor Sohal is Asia Pacific Editor of the International Journal of Qual-
ity and Reliability Management and Editorial Board member of a number
of journals in the area of quality management, technology management
and operations management. He has authored or co-authored over 60
research papers published in refereed journals, as well as three books and
a number of chapters contributed to books. His current research interests
are in manufacturing/operations, strategy, technology management, quality
management and just-in-time production systems. He has received
research grants from the State and Federal Governments, the Australian
Research Council and Monash University.