Journal of Operations Management 19 (2001) 541–558
A taxonomy of manufacturing strategies revisited Markham T. Frohlich a,∗ , J. Robb Dixon b,1 a
Operations and Technology Management Department, London Business School, Regents Park, London NW1 4SA, UK b Operations Management Department, Boston University School of Management, Boston, MA 02215, USA Received 18 June 1999; accepted 16 April 2001
Abstract While our field has done commendable work putting forward new ideas in operations strategy, we have historically done a less effective job validating concepts after their introduction. Given this issue, we attempted to test and extend one of the most influential OM configurations — Miller and Roth’s [Management Science 40 (1994) 285] taxonomy of manufacturing strategies. Their taxonomy was longitudinally replicated with a newer set of North America Manufacturing Futures data as well as an entirely different global sample. Our replications partially supported Miller and Roth’s taxonomy of three strategy-types (Caretakers, Marketeers, and Innovators), but found no evidence of the two underlying dimensions of manufacturing strategy that they called market scope and differentiation. Interestingly, the Marketeers were replaced in the 1990s by a new strategy called Designers, and three other unique manufacturing strategies were identified in the global data. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Operations strategy; Taxonomies; Empirical research
1. Introduction In the late 1980s, Miller and Roth identified three manufacturing strategies — Marketeers, Caretakers, or Innovators — commonly used by North American manufacturers. Miller and Roth’s (1994) taxonomy has since become one of the most influential frameworks in the manufacturing strategy literature, and a check of leading OM journals (Goh et al., 1997) shows their work cited by over 40 subsequent studies. The Marketeers’ strategy is oriented towards reliability in the manufacturing process — especially in quality and delivery. The Caretakers’ strategy is uniquely preoc∗
Corresponding author. Tel.: +44-207-262-5050; fax +44-207-724-7875. E-mail addresses:
[email protected] (M.T. Frohlich),
[email protected] (J.R. Dixon). 1 Tel.: +1-617-353-5243; fax: +1-617-353-4098.
cupied with low price over all other potential competitive capabilities. The Innovators’ strategy is characterized by an emphasis on quality and an avoidance of price competition. Miller and Roth’s taxonomy, combined with the emerging theory that top–down alignment of business strategy and competitive capabilities drives performance (Hayes and Wheelwright, 1984; Hill, 1995; Flynn et al., 1999), form two important cornerstones of our existing knowledge about the content of manufacturing strategy (Fig. 1). While we conceptualize manufacturing strategy in relatively static taxonomic terms, manufacturing and its strategic role seem to have steadily changed since Miller and Roth’s, 1987 study. In an accelerating trend throughout the 1980’s, many manufacturers grappled with quality and flexibility issues and adopted new approaches like SPC, JIT, and CAD/CAM to improve their competitiveness. Triggered by the recession in the early 1990s, manufacturers next reengineered their
0272-6963/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 2 7 2 - 6 9 6 3 ( 0 1 ) 0 0 0 6 3 - 8
542
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Fig. 1. Competitive capabilities and manufacturing strategy types.
business processes and implemented ERP systems in an attempt to reduce costs and improve service. Beginning in the mid-1990s, growing global competition and a booming economy led many manufacturers to focus on improved delivery in their supply chains as well as price, new product development, and service. It has even led many formerly outstanding manufacturers to rely on outsourcing, thereby abandoning their own operations. There also seem to be some
underlying differences, perhaps due to national culture, between manufacturers in various regions of the emerging global economy. These issues all raise important questions concerning our present knowledge about manufacturing strategy. Is Miller and Roth’s taxonomy still applicable in today’s dynamic global manufacturing environment? If not, is their taxonomy baseless, or can new findings update their model into a more robust framework?
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
2. The role of replication in theory building Theoretical discovery is not the final goal of research, and the principle of replication is widely recognized as the hallmark of scientific knowledge (Rosenthal and Rosnow, 1984; Blaug, 1992). Theory building must be followed by rounds of verification and elaboration (Flynn et al., 1990). Kuhn (1962) called this testing of existing theory “normal science”. It consumes the long periods between paradigm shifts and accounts for a significant portion of the knowledge developed by a discipline. In other fields, important findings are therefore regularly replicated at first on purpose and later as a routine research task (Lindsay and Ehrenberg, 1993). Supporting this point, the survey of science editors by Madden et al. (1995) found that they typically endorsed replication as a necessary part of research. Cumulative research is especially important because theories can never be entirely proved or disproved (Hempel, 1966). Each successful duplication simply provides greater support for a theory while every failed replication raises further concerns (Popper, 1978). Authors from a host of management fields have noted similar concerns with unreplicated ideas and the need for greater verification (Dewald et al., 1986; Hubbard and Armstrong, 1994; Lindsay, 1994; Hubbard and Vetter, 1996; Hubbard et al., 1998). Researchers’ biases inevitably affect how observations are gathered and interpreted, and the only way to avoid this is to replicate research (Judd et al., 1991). As Hubbard et al. (1998) argued, a strategy literature dominated by unreplicated findings is of marginal value and especially needs a balance between the publication of novel results and replication studies. Without replication our hopes for developing a consensus regarding manufacturing strategy and cumulative knowledge will continue to elude us (Amundson, 1998). The cornerstone of science is that no view is immune to challenge (Selltiz et al., 1976), and as every field matures weak models and theories need to be abandoned and the surviving ones continually refined. “Replication allows us to demonstrate that something is really there, with extensions to see how broadly and in what instances it exists and can be used” (Hubbard et al., 1998). Nevertheless, replications are still rare in the OM literature. Part of the reluctance to refute
543
previous work is perhaps the reputations of the scholars that developed these models. Kane (1984) pointed out that conducting replications can also be misconstrued as mediocrity, or bullying, and may therefore be intentionally avoided. Having said that, physicists have no trouble challenging and extending the ideas of Einstein, psychologists of Freud, biologists of Darwin, or economists of Keynes. This is what normal science is all about (Kuhn, 1962), and for our field to truly mature we need to routinely assume more of these responsibilities. Towards that end, we replicated Miller and Roth’s (1994) taxonomy of manufacturing strategies using data that are more recent. The contribution of this paper lies in not only longitudinally and globally testing Miller and Roth’s ideas, but also in serving as a guide for others interested in replicating OM models and theories.
3. A taxonomy of manufacturing strategies and study hypotheses One of the most popular approaches to strategy research is the identification of strategic configurations: groups of organizations that mirror each other based upon some logical subset of variables (Meyer et al., 1993; Bozarth and McDermott, 1998). The assumption underlying all configuration research is that taxonomy classifications are a meaningful way to capture the complexities of organizational reality (Dess et al., 1993; Ketchen and Shook, 1996). While not theories per se, configurations are a critical part of theory development (Bacharach, 1989). Taxonomies are especially valuable for initially classifying data before subsequent analyses (Galbraith and Schendel, 1983; Ketchen et al., 1993). To review Miller and Roth’s (1994) taxonomy, their model was developed using a cluster analysis of the relative importance that 188 North American manufacturers in 1987 attached to 11 competitive capabilities which define the manufacturing task. The study also uncovered by canonical discriminant analysis two main underlying dimensions along which manufacturers in each cluster differed. One dimension concerned the ability of firms to differentiate themselves from their competition (market differentiation) while the other described the scope of their product lines and markets (market scope).
544
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
3.1. North American replication hypotheses The first set of hypotheses builds on Miller and Roth’s (1994) original 1987 findings and tests their taxonomy’s longitudinal validity for North American manufacturers in two different data sets. As Miller and Roth (1994) commented, “an important line of future research is to test the stability of this taxonomy globally, and over time.” In order to test Miller and Roth’s North American model, four hypotheses were formulated as follows: H1a. The 1994 data will yield three clusters that reflect the same competitive capability rankings Miller and Roth found in 1987 for North American manufacturers. H1b. The 1998 data will yield three clusters that reflect the same competitive capability rankings Miller and Roth found in 1987 for North American manufacturers. H2a. The 1994 competitive capabilities will form the two market differentiation and scope dimensions that Miller and Roth found in 1987 for North American manufacturers. H2b. The 1998 competitive capabilities will form the two market differentiation and scope dimensions that Miller and Roth found in 1987 for North American manufacturers. 3.2. Global extension hypotheses As noted above, the Innovators, Marketeers, and Caretakers strategic groups were identified based upon a sample of North American manufacturers. An obvious extension of Miller and Roth’s work is to test their framework around the world. Accordingly, we also formulated three additional hypotheses for manufacturing strategy in a global context as follows: H3. An analysis of South American, Western European, and Asia-Pacific manufacturers will yield Caretakers, Marketeers, and Innovators just as Miller and Roth’s analysis of North American manufacturers did.
H4. The Caretakers, Marketeers, and Innovators strategic clusters for South American, Western European, and Asia-Pacific manufacturers will reflect the same competitive capability rankings that North American manufacturers’ clusters did. H5. The competitive capabilities for South American, Western European, and Asia-Pacific manufacturers will form the same underlying dimensions of manufacturing strategy that North American manufacturers’ competitive capabilities did.
4. Research methods Data from the 1994 Boston University Manufacturing Futures Project (MFP) Survey and the 1998 International Manufacturing Strategy Survey (IMSS) were used to replicate Miller and Roth’s study. In 1994, the MFP responses were collected from 212 business units in the same machinery, electrical, basic and industrial, and consumer industries included in Miller and Roth’s 1987 sample. The response rate for 1994 was 21 percent, similar to that of the 1987 survey. The overlap between the 1987 and 1994 samples consisted of 12 business units. The IMSS was founded in 1992 to gather data related to manufacturing strategy in a global setting (see Voss and Blackmon, 1998, or Frohlich and Westbrook, 2001, for more IMSS details). The survey focused on the ISIC Division 38: Manufacture of Fabricated Metal Products, Machinery and Equipment. For 1998, researchers in 23 countries administered the survey and data were collected from 703 companies (Table 1). The response rate for the total sample was approximately 20 percent. In nations where English was not a prevalent language the surveys were translated by full-time OM professors familiar with manufacturing strategy. With only two exceptions, the questions asked in the 1987 survey were duplicated in 1994 and 1998. As seen in Table 2, 11 questions were asked in 1987 on the competitive capabilities that manufacturers emphasized. In 1994, the MFP question concerning the capability to advertise and promote products was dropped based upon feedback from managers. Similarly, the 1998 IMSS study did not ask about broad distribution. The 1994 MFP data were measured on the same 1–7 Likert scale as the 1987 survey (“1 = least important”
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
545
Table 1 The 1998 IMSS sample by region and country North America
South America
Western Europe
Asia-Pacific
Canada (39) Mexico (29) United States (41)
Argentina (31) Brazil (27) Chile (10) Peru (8)
Denmark (27) Finland (14) Germany (29) Hungary (38) Italy (71) Netherlands (29) Norway (13) Spain (33) Sweden (27) United Kingdom (24)
Australia (55) China (30) Hong Kong (14) Japan (29) New Zealand (32) South Korea (50)
to “7 = most important”). The 1998 IMSS data used a similar 1–5 Likert scale. Except for these deviations, the 1994 and 1998 measures duplicated Miller and Roth’s work. A refinement of the statistical technique used by Miller and Roth (1994) in their original analysis was employed to test our hypotheses. Miller and Roth relied on only a non-hierarchical algorithm (also know as the K-means or iterative method) for their cluster analysis. More recently, experts strongly advocate using a two-step clustering procedure whereby a hierarchical method first determines the number of clusters and cluster centroids which are then inputs into the subsequent non-hierarchical algorithm (Ketchen and Shook, 1996). We followed this preferred two-step approach in our study. Ward’s partitioning and squared Euclidean distance were used for the hierarchical cluster analysis. Partitioning using Ward’s method was selected because of its established
robustness, ability to maximize within-cluster homogeneity and between-cluster heterogeneity, and capability to recover known cluster structure (Aldenderfer and Blashfield, 1984). The squared Euclidean distance measure is recommended with Ward’s method because it leads to clusters with the smallest sum of squares error (Arabie and Huber, 1994). 5. Research results 5.1. Hypothesis tests for the 1994 and 1998 North American data The most appropriate number of clusters for the 1994 and 1998 North American data were determined using multiple methods (Ketchen and Shook, 1996). First, the dendogram for the cluster analysis was visually inspected for relatively dense “branches” to
Table 2 Competitive capability questions in the 1987, 1994 and 1998 surveys No.
Competitive capabilities
MFP 1987
MFP 1994
IMSS 1998
1 2 3 4 5 6 7 8 9 10 11
Low price Design flexibility Broad product line Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad distribution Advertising
X X X X X X X X X X X
X X X X X X X X X X Dropped in 1994
X X X X X X X X X Not asked Not asked
546
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
confirm the number of major groups that had been formed (Aldenderfer and Blashfield, 1984). Second, the incremental changes in the agglomeration coefficient were examined. Hierarchical cluster analysis builds a decision tree or pyramid-like structure by adding (i.e. agglomerating) individual elements one at a time. Relatively large changes in the agglomeration coefficient at the later cluster combination stages signify important increases in cluster homogeneity. A relatively large increase in the agglomeration coefficient implies that dissimilar clusters have been merged together at that step, and therefore the number of clusters just before the merger is the most appropriate choice (Ketchen and Shook, 1996). To validate the three-group model, we tested for heterogeneity across clusters based upon the competitive capabilities. All 10 competitive capabilities differed significantly in a test of mean rank differences across the three clusters (P < 0.000) which suggested that distinct groups were identified. A split-half analysis was next conducted to assess the reliability and stability of the proposed three-cluster solution. We selected a six-cluster solution as the starting point for the split-half analysis and tested six-, five-, four-, three- and two-cluster models. In this analysis, the three-cluster solutions in each of the two split-halves shared the most uniform similarities. Our decision to employ a three-cluster solution was likewise based on interpretability. More specifically, moving from a four- to three-cluster solution combined two similar clusters, whereas moving from a threeto two-cluster solution forced together two dissimilar clusters. Thus, we concluded that a three-cluster solution was most appropriate given the reliability across split-halves and the overall interpretability of the clusters. These three clusters, similar in relative size to the 1987 sample, are shown in Table 3. Hypotheses H1a and H1b held that cluster groups would reflect the same competitive capabilities in 1994 and 1998 as they did in 1987. The results are presented in the Tables 4–6. The mean scores are shown with their overall ranks in parentheses. Spearman’s rank–order correlation coefficients (rs ) were calculated to compare the competitive priority rankings between 1987 and 1994, 1987 and 1998 and 1994 and 1998 for each pair of clusters. This is reported at the bottom of each table, along with its significance level. H1a was supported for Caretakers, Marketeers, and
Table 3 The three clusters in 1987, 1994 and 1998 for North America dataa Clusters Caretakers Marketeers Innovators
MFP 1987 (%)
MFP 1994 (%)
IMSS 1998 (%)
11 (18) 49 (81) 40 (65)
20 (42) 48 (100) 31 (65)
26 (26) 43 (43) 31 (31)
100 (164)b
100 (207)c
100 (100)c
a
Numbers in parentheses equals the actual count. SAS AceCLUS and FastCLUS excludes cases with missing data and eliminates statistical outliers, thus the 1987 total of 188 cases was reduced to 164. c SPSS K-means listwise deletion for missing data reduced the 1994 tally from 212 to 207 cases and the 1998 total of 109 surveys to 100 cases. b
Innovators. H1b was partially supported for Caretakers and Innovators, but there was no trace of the Marketeers in the 1998 data. Following Miller and Roth’s original study, and as a final test of Hypotheses H1a and H1b, discriminant analysis was run to ensure that the three-cluster groups were correctly classified. The competitive capabilities listed in Table 2 were entered as independent variables and cluster membership (Caretakers, Marketeers, or Innovators) as the dependent variable. As seen in Table 7, the cluster algorithms correctly classified the majority of cases for 1994 and 1998. The discriminant Table 4 Comparison of Caretakers’ mean competitive capabilities for North America dataa Competitive capabilities
MFP 1987b
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line Broad distribution
6.06 5.11 4.17 5.83 4.17 5.55 6.00 2.94 5.00 4.78
(1)d (5) (8) (3) (8) (4) (2) (10) (6) (7)
MFP 1994b
IMSS 1998c
6.17 4.02 4.50 5.90 4.52 4.90 5.81 2.81 4.48 3.93
4.39 (1) 1.92 (9) 2.73 (7) 4.31 (2) 3.69 (5) 3.77 (4) 3.88 (3) 3.46 (6) 2.15 (8) N.A.
(1) (8) (6) (2) (5) (4) (3) (10) (7) (9)
MFP 1987 to MFP 1994 Spearman’s rs = 0.79, significant at 0.007 level; MFP 1987 to IMSS 1998 Spearman’s rs = 0.69, significant at 0.041 level; MFP 1994 to IMSS 1998 Spearman’s rs = 0.90, significant at 0.001 level. b Measured on a 1–7 Likert scale. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores. a
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
547
Table 5 Comparison of Marketeers’ mean competitive capabilities for North America dataa
Table 6 Comparison of Innovators’ mean competitive capabilities for North America dataa
Competitive capabilities
MFP 1987b
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line Broad distribution
5.70 5.07 5.04 6.78 6.07 5.59 6.11 5.27 5.30 5.68
(4)d (9) (10) (1) (3) (6) (2) (8) (7) (5)
MFP 1994b
IMSS 1998c
Competitive capabilities
MFP 1987b
6.46 5.72 5.57 6.64 5.98 6.30 6.53 5.99 5.95 6.04
4.02 (8) 4.14 (7) 3.91 (9) 4.51 (3) 4.58 (1) 4.44 (5) 4.51 (3) 4.58 (1) 4.16 (6) N.A.
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line Broad distribution
5.46 5.98 4.06 6.75 6.38 5.40 6.29 5.52 4.78 3.92
(3) (9) (10) (1) (7) (4) (2) (6) (8) (5)
(6)d (4) (9) (1) (2) (7) (3) (5) (8) (10)
MFP 1994b
IMSS 1998c
5.37 5.58 5.06 6.60 5.62 5.33 6.15 5.69 5.42 5.33
2.77 (9) 3.13 (6) 2.81 (8) 4.71 (2) 4.83 (1) 4.00 (5) 4.23 (4) 4.61 (3) 2.94 (7) N.A.
(7) (5) (10) (1) (4) (8) (2) (3) (6) (8)
a MFP 1987 to MFP 1994 Spearman’s r = 0.84, significant s at 0.002 level; MFP 1987 to IMSS 1998 Spearman’s rs = 0.42, significant at 0.260 level; MFP 1994 to IMSS 1998 Spearman’s rs = 0.51, significant at 0.158 level. b Measured on a 1–7 Likert scale. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores.
a MFP 1987 to MFP 1994 Spearman’s r = 0.83, significant s at 0.003 level; MFP 1987 to IMSS 1998 Spearman’s rs = 0.78, significant at 0.013 level; MFP 1994 to IMSS 1998 Spearman’s rs = 0.68, significant at 0.042 level. b Measured on a 1–7 Likert scale. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores.
analyses indicated extremely good differentiation among the three groups and was substantially above the 33% correct classification expected by chance. Hypotheses H2a and H2b predicted that the 1994 and 1998 competitive capabilities would form the two underlying dimensions called market differentiation and market scope that Miller and Roth found in 1987. Miller and Roth used canonical discriminant analysis in their original study to reach this conclusion. A similar canonical discriminant analysis procedure was run using the 1994 and 1998 data. Canonical discriminant analysis derives a linear combinations of the competitive capabilities that have the highest possible multiple correlation with the clusters. As with Miller and Roth’s 1987 results, two canonical functions were statistically significant for the 1994 and 1998 data. The results are presented in Table 8.
Table 9 shows the canonical loadings for Functions 1 and 2 in 1987, 1994 and 1998. The canonical loadings are interpreted similarly to factor loadings in principal component analysis, and represent the correlations between the original competitive capabilities and an underlying dimension. Table 9 essentially shows that each cluster favors the underlying dimensions to different degrees. For example, looking at Tables 4–6 for 1987, if a cluster placed a greater priority on conformance quality, performance quality, and after-sales service, then it will tend to fall at the high end (positive side) of market differentiation (Function 1). Conversely, if in 1987 the cluster placed less priority on conformance, performance, and service it will fall on the lower side (negative) of the market differentiation dimension. As seen in Table 9, there was no support for either H2a or H2b (market
Table 7 Results of discriminant analysis for 1994 MFP and 1998 IMSS North America data Breakout of items correctly/incorrectly classified Caretakers 95% Caretakers Marketeers Innovators
Marketeers 95%
1994 MFP percent correctly classified 40 2 0 95 0 2
Innovators 97%
Caretakers 96%
Marketeers 93%
0 5 63
1998 IMSS percent correctly classified 25 1 1 40 0 1
Innovators 97% 0 2 30
548
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Table 8 Canonical discriminant functions for 1987, 1994 and 1998 for North America data Canonical function
Canonical correlation
Significance of function
MFP 1987 1 2
0.76 0.74
0.000 0.000
MFP 1994 1 2
0.88 0.56
0.000 0.000
IMSS 1998 1 2
0.84 0.71
0.000 0.000
differentiation and market scope), so we renamed canonical functions 1 and 2 for 1994 and 1998 based upon our findings. 5.2. Hypothesis tests for the 1998 global data A set of tests similar to the above analyses was repeated using the 1998 IMSS global data. Hypo-
thesis H3 predicted that the three North American strategic clusters would reappear in the global data. For the 1998 South American data, three clusters provided the best fit based upon the dendogram, incremental changes in the agglomeration coefficient, split-half analysis, and interpretation of the clusters. These three clusters are shown in Table 10. Identical analyses were done using the Western Europe and Asia-Pacific sub-samples, and the results are summarized in Table 10. Surprisingly, four clusters emerged in both the Western Europe and Asia-Pacific sub-samples. Again, discriminant analysis was run and the cluster algorithm correctly classified the great majority of the 1998 IMSS global cases (bottom of Table 10). H3 was partially supported. Caretakers and Innovators were found along with three new clusters: Idlers, Servers, and Mass Customizers. Hypothesis H4 held that global cluster groups will reflect the same competitive capabilities in 1998 as their North American counterparts did. The results are shown in the Tables 11–14. Spearman’s rank–order correlation coefficients (rs ) were also calculated to compare the competitive priority rank-
Table 9 Canonical loadings (correlations) and means for 1987, 1994 and 1998 North America dataa differentiation
1987 Market service
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line Broad distribution
Function 1 −0.164 0.176 0.061 0.698 0.799 −0.035 0.122 0.653 −0.014 −0.067
1994 After-sales design 0.113 0.323 0.232 0.204 0.271 0.302 0.211 0.532 0.317 0.380
1998 Product scope
1987 Market price
1994 Low price
1998 Low
−0.052 0.744 0.305 0.311 0.067 0.223 0.200 0.367 0.751 N.A.b
Function 2 0.097 −0.415 0.435 0.134 −0.063 0.109 −0.131 −0.004 0.239 0.666
0.523 −0.132 0.467 −0.122 −0.244 0.053 0.163 −0.451 0.456 0.216
0.727 −0.023 0.239 −0.212 −0.408 0.071 −0.006 −0.350 0.181 N.A.b
Cluster means MFP 1994 Marketeers Caretakers Innovators Designers a b
Function 1 1.703 −3.078 −0.632 –
IMSS 1998 Function 2 0.334 0.758 −1.003 –
Function 1 – −2.270 −0.254 1.556
Note: canonical loadings greater than 0.40 are significant and are in bold print. N.A.: not applicable.
Function 2 – 0.825 −1.462 0.555
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
549
Table 10 Clusters in 1998 for the IMSS global dataa Clusters
South America
Western Europe
Asia-Pacific
Caretakers Designers Innovators Idlers Servers Mass Customizers
21% (14) 73% (48) Not found 6% (4) Not found Not found 100% (66)
19% (53) 52% (151) 4% (10) Not found 25% (72) Not found 100% (286)b
31% (55) 62% (112) 11% (20) Not found Not found 4% (8) 100% (181)c
Designers 98%
Innovators N.A.
Idlers 100%
Designers 95%
Innovators 100%
Servers 88%
Designers 91%
Innovators 100%
Mass Customizers 100%
Results of discriminant analysis South America Caretakers 97% Western Europe Caretakers 94% Asia-Pacific Caretakers 93% a
Numbers in parentheses equals the actual count. SPSS K-means listwise deletion for missing data reduced the 1998 total of 286 surveys to 266 cases. c SPSS K-means listwise deletion for missing data reduced the 1998 total of 195 surveys to 181 cases. b
ings between each pair of North American and global strategic clusters. H4 was partially supported. The Caretakers were found globally and the Innovators showed up in Western Europe and AsiaPacific. H5 predicted that the 1998 global sub-sample would form the same underlying dimensions as the North American data. The results are presented in Table 15. Tables 16 and 17 shows the canonical loadings and
the cluster means for the functions based on the global data, and there was no support for H5. 6. Discussion 6.1. Hypothesis 1 Hypothesis 1, that the three North American groups will reflect the same competitive capabilities in 1994
Table 11 Comparison of Caretakers’ mean capabilities for the 1998 IMSS global dataa No.
Competitive capabilities
1998 IMSS North Americab
1 2 3 4 5 6 7 8 9
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line
4.39 1.92 2.73 4.31 3.69 3.77 3.88 3.46 2.15
(1)d (9) (7) (2) (5) (4) (3) (6) (8)
1998 IMSS South Americac
1998 IMSS Western Europec
1998 IMSS Asia-Pacificc
4.50 1.57 2.64 4.71 3.92 3.71 4.14 3.21 2.00
4.07 2.58 3.73 3.66 3.43 3.98 4.05 2.86 3.35
4.27 2.67 2.61 4.12 3.70 4.32 3.96 4.09 2.74
(2) (9) (8) (1) (4) (5) (3) (6) (7)
(1) (9) (4) (5) (6) (3) (2) (8) (7)
(2) (8) (9) (3) (6) (1) (5) (4) (7)
a The 1998 IMSS North America to South America Spearman’s r = 0.95, significant at 0.000 level; 1998 IMSS North America s to Western Europe Spearman’s rs = 0.78, significant at 0.013 level; 1998 IMSS North America to Asia-Pacific Spearman’s rs = 0.78, significant at 0.013 level. b Repeated from Table 4. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores.
550
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Table 12 Comparison of designers’ mean capabilities for the 1998 IMSS global dataa No.
Capabilities
1998 IMSS North Americab
1 2 3 4 5 6 7 8 9
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line
4.02 4.14 3.91 4.51 4.58 4.44 4.58 4.58 4.16
(8)d (7) (9) (3) (1) (5) (3) (1) (6)
1998 IMSS South Americac
1998 IMSS Western Europec
1998 IMSS Asia-Pacificc
4.12 3.66 3.89 4.25 4.35 4.29 4.16 4.25 3.77
3.50 3.82 3.80 4.48 4.54 4.18 4.42 4.17 3.74
3.92 3.98 3.86 4.71 4.72 4.40 4.22 4.53 4.19
(6) (9) (7) (3) (1) (2) (5) (3) (8)
(9) (6) (8) (2) (1) (4) (3) (5) (7)
(8) (7) (9) (2) (1) (4) (5) (3) (6)
The 1998 IMSS North America to South America Spearman’s rs = 0.73, significant at 0.024 level; 1998 IMSS North America to Western Europe Spearman’s rs = 0.83, significant at 0.005 level; 1998 IMSS North America to Asia-Pacific Spearman’s rs = 0.93, significant at 0.000 level. b Repeated from Table 5. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores. a
(H1a) and 1998 (H1b) that they did in 1987, was only partially supported. The Caretakers and Innovators have maintained relative stability across all competitive capabilities. The Marketeers showed a different trend; they have remained consistent on the two critical dimensions of conformance quality and delivery dependability but have fluctuated on most other variables. As seen in Tables 3–6, there was strong evidence that while the Innovators and Caretakers endured over a decade, the Marketeers’ approach was replaced in the 1990s by a new manufacturing strategy. While cluster analysis cannot prove these were (or were not) the
originally identified Caretaker, Marketeer, and Innovator strategies, the significant Spearman correlations firmly suggest it. From 1987 to 1998, the Caretakers seem to have grown in relative proportion at the expense of the Innovators. The third group has stayed more constant in terms of relative size. These three manufacturing strategies each merit further discussion. 6.1.1. Miller and Roth’s Caretakers Miller and Roth (1994) noted three distinguishing characteristics of Caretakers. First, low price was their dominant competitive capability. Second, delivery
Table 13 Comparison of Innovators’ mean capabilities for the 1998 IMSS global dataa No.
Capabilities
1998 IMSS North Americab
1 2 3 4 5 6 7 8 9
Low price Design flexibility Volume flexibility Conformance quality Performance quality Delivery speed Delivery dependability After-sales service Broad product line
2.77 3.13 2.81 4.71 4.83 4.00 4.23 4.61 2.94
(9)d (6) (8) (2) (1) (5) (4) (3) (7)
1998 IMSS South Americac
1998 IMSS Western Europec
1998 IMSS Asia-Pacificc
Not Not Not Not Not Not Not Not Not
2.20 3.10 1.40 4.40 4.80 4.18 2.70 3.60 2.80
2.25 2.75 3.75 4.75 4.70 4.00 4.15 4.25 2.60
found found found found found found found found found
(8) (5) (9) (2) (1) (3) (7) (4) (6)
(9) (7) (6) (1) (2) (5) (4) (3) (8)
The 1998 IMSS North America to Western Europe Spearman’s rs = 0.85, significant at 0.004 level; 1998 IMSS North America to Asia-Pacific Spearman’s rs = 0.93, significant at 0.000 level. b Repeated from Table 6. c Measured on a 1–5 Likert scale. d Numbers in parentheses are ranks based on mean scores. a
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
551
552
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Table 15 Canonical discriminant functions for 1998 IMSS global data Canonical function
Canonical correlation
Significance of function
South America 1 2
0.82 0.78
0.000 0.000
Western Europe 1 2 3
0.80 0.71 0.44
0.000 0.000 0.000
Asia-Pacific 1 2 3
0.82 0.69 0.57
0.000 0.000 0.000
dependability, delivery speed, and conformance quality were relatively high priorities. Finally, their least important concerns were after-sales service and high performance products. Table 4 suggests that the North American Caretakers were consistent in these competitive capabilities between 1987 and 1998. Low price remained the top priority in 1994 and 1998, while conformance quality, delivery speed, and delivery dependability interchangeably swapped second, third, or fourth places between the three surveys. Table 9 also supports the argument that there are unique characteristics to the unidimensional or non-differentiated Caretakers’ strategy. In 1994, the Caretakers had the lowest emphasis (−3.078) on Function 1 (after-sales service) and the highest mean score (0.758) on Function 2 (low price). This perfectly fits the Caretakers’ profile. The same pattern occurred in 1998 with a −2.270 for Function 1 (product design) and 0.825 for Function 2 (low price). 6.1.2. Miller and Roth’s Innovators Miller and Roth determined that Innovators as a group shared certain characteristics with Marketeers. In particular, this cluster most highly valued conformance and performance quality along with delivery dependability. Unlike the Marketeers and Caretakers, however, price was a comparatively lower priority for Innovators. Table 6 verifies the 1987 findings. In 1994 and 1998 conformance quality, performance quality, and delivery dependability remained among the Innovators’ top concerns. After-sales service also
moved up to become an important priority in 1998. Delivery speed has recently emerged as an important issue for Innovators in 1998, while low price and volume flexibility have consistently stayed low priorities. In short, the Innovators appeared to stay relatively stable in their competitive thrusts for the past decade. Performance and conformance quality remained top concerns, while price has gradually dropped in importance. Like the Caretakers, the canonical loadings in Table 9 confirm the nature of the Innovators’ strategy. Given their focus on conformance and performance quality, they scored low in 1994 for Function 1 (after-sales service −0.632) and the absolute lowest for Function 2 (low price −1.003). A similar pattern occurred in 1998. 6.1.3. Miller and Roth’s Marketeers Miller and Roth found in their 1987 study that the top priorities of Marketeers were conformance quality, dependable deliveries, and performance quality. They further noted that Marketeers demonstrated some price consciousness. Table 5 shows that this enigmatic third group has reshuffled its top priorities between 1987 and 1998. In 1987, conformance quality stayed number one in 1994 but dropped to third by 1998. In 1994, price moved up to third but by 1998 was a much lower priority. In 1998, performance quality and after-sales service became the top priorities of this cluster closely followed by conformance quality and delivery dependability. Based on our evidence, Miller and Roth’s, 1987 Marketeer strategy no longer exists. Instead, it was replaced in the mid-1990s by a new manufacturing strategy that we call the Designers. Important clues to the identity of this new third cluster reside in the tables. Two of the biggest differences between the three manufacturing strategy types were in the high degrees of emphasis placed by the new group on design flexibility (4.14) and broad product lines (4.16) in comparison to the 1998 Caretakers and Innovators (see Tables 4–6). Consequently, this new third cluster placed the only positive emphasis (1.556) on Function 1 that reflected product design capabilities (Table 9). It is also worth noting that this third cluster, unlike Caretakers and Innovators, consistently competed on two differentiated fronts. Just as the Marketeers excelled on both the market differentiation and market
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
553
554
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Table 17 Cluster means of canonical loadings for 1998 IMSS global data Function 1
Function 2
Function 3
IMSS 1998 Western Europe Designers 1.215 Caretakers −0.708 Innovators −1.848 Servers −1.771
0.243 −1.913 2.389 0.567
0.036 −0.315 −2.174 0.459
IMSS 1998 Asia-Pacific Designers 1.097 Caretakers −1.910 Innovators 0.062 Mass Customizers −2.371
0.319 0.288 −2.702 0.298
−0.033 0.523 −0.006 −3.115
IMSS 1998 South America Designers 0.289 Caretakers 0.576 Idlers −5.493
0.708 −2.311 −0.401
scope dimensions in the 1980s, the Designers tried to compete with the twin weapons of product development and low price (0.555 for Function 2) in the 1990s (Table 9). In summary, this cluster shifted its competitive capabilities over time, although the group’s relative size stayed constant. An important finding from this study is that at least one strategic group does not dogmatically compete (like the Caretakers and Innovators do) on the same unidimensional sets of competitive capabilities, but instead changes based upon the business environment. Perhaps driven by the early 1990s’ recession, and next the boom beginning in 1995, this group’s major competitive capabilities have seesawed over time. Important distinguishing characteristics of Designers are a dual emphasis on performance quality and after-sales service along with an accent on new product design. Like the old Marketeers, the new Designers also emphasize conformance quality and delivery dependability. Similarly, the Designers approach manufacturing strategy from a differentiated perspective and try to excel on multiple fronts. 6.2. Hypothesis 2 In contrast to Hypotheses H1a and H1b, the findings from this study lead to the rejection of Hypotheses H2a and H2b. In 1994 and 1998 the competitive capabilities did not form the market differentiation and scope
dimensions that they once did in 1987. The reason for this vacillation in underlying dimensions is mainly due to the Marketeers’ and later the Designers’ fluctuating competitive capabilities. As the Marketeers and Designers maneuvered for competitive advantage, they shifted the underlying playing field for all manufacturers. Thus, Miller and Roth’s first canonical function (market differentiation) and Function 2 (market scope) repeatedly changed over the past decade. 6.3. Hypotheses 3, 4 and 5 H3 and H4 predicted that the Caretakers, Marketeer, and Innovator strategic groups would be found in the global data and would emphasize competitive capabilities similar to those of the North American samples. There was only partial support for these hypotheses. The Caretakers strategic group was found in the South American, Western European, and Asia-Pacific sub-samples (Table 11). Our findings suggest that the Caretakers approach is the only manufacturing strategy universal across time and location. The new Designers strategy was found in all three sub-samples (Table 12) while the Innovators were identified in the Western European and Asia-Pacific data (Table 13). Perhaps the biggest surprise was the relative size of each group around the world (Table 10). Notably, the Innovators strategy type was relatively uncommon in Western Europe and Asia-Pacific, while the Caretakers and Designers tended to make up larger percentages of the sub-samples than in North America. An unexpected finding was three additional strategy types — one each in South American, Western Europe and Asia-Pacific. As shown in Tables 10 and 14, a small percentage of South American companies (6%) reported very little emphasis on any of the competitive capabilities. We named this cluster the Idlers because they seemed to lag behind the typical company in aggressively emphasizing some combination of competitive capabilities. A relatively large cluster (25%) emerged in the Western European data oriented around after-sales service. We accordingly named this group the Servers. Finally, a relatively small number of Asia-Pacific companies (4%) were simultaneously most interested in low price along with design flexibility. This cluster seemed to fit the profile of what is popularly called mass customization (Pine, 1993), so we labeled this strategic group the Mass Customizers.
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
The literature has long hinted at Europe’s Servers and Japan’s Mass Customizers strategies, but this is the first time that these two hypothetical groups have been confirmed in a formal study. The literature notes Europe’s unique service culture (Gronroos, 1983; Gummesson, 1991) while European products from German cars to Swiss watches are legendary for their after-sales customer support. In many ways, the focused Servers’ strategy mimics the Innovators in that they emphasized a single capability that appeals to customers (i.e. after-sales service) while they downplayed other capabilities like price and flexibility. Similarly, most examples of Mass Customizers in the 1990s’ popular press were of Japanese manufacturers (see for example, Gilmore and Pine, 1997), which parallels our finding of this manufacturing strategy only in the Asia-Pacific data. Conceptually, the Mass Customizers are also like the Innovators and Servers in that they focused their manufacturing strategy around a relatively narrow set of competitive capabilities (e.g. low price and design flexibility) to the exclusion of other dimensions. H5 was rejected. As with H2a and H2b, there was no evidence that constant dimensions underlie competitive capabilities globally (Table 16). Miller and Roth possibly misinterpreted their canonical functions as the underlying dimensions that Caretakers, Marketeers, and Innovators compete upon. An alternative hypothesis is that these underlying dimensions reflect what manufacturers in different parts of the world compete upon. Given that possibility, by the late 1990s it appeared that North American manufacturers were most interested in product design and low price (Functions 1 and 2 in Table 9 for 1998 IMSS data). South American companies concentrated on low price and design flexibility (Functions 1 and 2 in Table 16). Western European manufacturers were preoccupied with manufacturing flexibility and after-sales service while Asia-Pacific manufacturers focused on quality and price (Table 16). 7. Conclusions 7.1. Theory replication As laid out in the beginning of the paper, this study attempted to replicate the taxonomy of manufacturing
555
strategy types put forward by Miller and Roth. Based on our findings, parts of their configuration were supported and others rejected for North American companies. Three clusters of strategy types appeared in each set of data, but the underlying dimensions along which the clusters differed in each survey shifted over time and differed across geographic regions. Consistent evidence emerged that two of Miller and Roth’s strategy types, Caretakers and Innovators, endured longitudinally from the 1980s through the 1990s with their core sets of competitive capabilities staying moderately to highly consistent. Meanwhile, the third, and largest, cluster appeared to change its competitive focus over time. One unanswered question is why there are three main manufacturing strategies. The theoretical explanation for Caretakers, Marketeers or Designers, and Innovators may rest in the links between Miller and Roth’s and Porter’s (1985) work. Porter argued that there are two sources of sustainable competitive advantage — low cost or differentiation — which lead to one of three generic business strategies: low price, differentiation, or focus. These generic business strategies coexist in an industry as companies battle for competitive advantage, which helps explain why multiple manufacturing strategies repeatedly showed up in both the North American and global data. According to Porter (1985), a company following a low price business strategy sets out to be a low cost producer by focusing on internal operations since on average 67% of all costs reside in manufacturing. Porter’s ideas, and the top–down OM theory of business and manufacturing strategy alignment (Fig. 1), therefore predict that if the business strategy is low price, then a manufacturer will relentlessly pursue low cost. This exactly parallels the Caretakers’ strategy. Porter’s work also explains why the Caretakers’ strategy is typically one of the smallest strategic groups, yet appears universally in the global data. In every industry there is room for only a few companies to exploit a low cost generic business strategy, but when successfully implemented, this becomes a potent source of competitive advantage. Conversely, companies embracing a differentiation business strategy seek to be unique along multiple dimensions that are widely valued by buyers (Porter, 1985). This closely fits the Marketeers’ approach in the 1980s and subsequently the Designers’ tactics in the 1990s. Porter argued that “the sustainability of
556
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
a differentiation strategy is usually greatest if differentiation stems from multiple sources, rather than resting on a single factor. . . ”. His theoretical argument closely parallels the empirical findings. Miller and Roth’s Marketeers’, and later our Designers, both aggressively pursued a wide variety of competitive capabilities in comparison to the Caretakers’ and Innovators. Notably, the Marketeers’ and Designers’ always attempted to dominate simultaneously the two underlying canonical dimensions of manufacturing strategy in the 1980s and 1990s (see Table 9). Porter’s ideas similarly clarify why the Marketeers’ strategy metamorphized during the 1990s into the Designers’ approach. A differentiation business strategy relies heavily on first recognizing what customers want and then choosing the best configuration of value-adding activities. Perhaps due to the maturation of electronics, communications, and information processing technologies, there was an explosion of interest in new product development (NPD) in the 1990s (Clark and Fujimoto, 1991; Rosenthal, 1992; Eisenhardt and Tabrizi, 1995). As increasingly sophisticated customers demanded new product offerings, companies began to emphasize NPD capabilities. In many sectors the life cycle or clockspeed (Fine, 1998) of products shortened with almost every generation. “Schumpeter’s ‘gale of creative destruction’, blowing at full force, fosters shorter product life cycles and rapid product obsolescence” (Schilling and Hill, 1998, p. 68). Therefore, NPD capabilities emerged in recent years as an important competitive capability for achieving the necessary degree of product differentiation (Kessler and Chakrabarti, 1996) and hence the Designers’ strategic approach. Finally, a focus business strategy means choosing a narrow competitive scope within an industry in which to compete (Porter, 1985). A focus strategy exploits the special needs of buyers in certain segments, given that these segments are poorly served by broadly-targeted competitors who serve them along with a host of other customers. “The essence of focus is the exploitation of a narrow target’s differences from the balance of the industry” (Porter, 1985, p. 15). This aptly describes the Innovators’ tenacious pursuit of conformance and performance quality, as well as the Servers’ concentration on service and the Mass Customizers’ convergence on flexibility. Moreover, Porter argued that there is often room for several sustainable
focus strategies in an industry, provided that the focusers choose different target segments which highlights why the Innovators’ strategy coexists with the Servers’ or Mass Customizers’ in the European and Asia-Pacific sub-samples. Interestingly, Porter (1985) also predicted that focus involves deliberately limiting potential sales volume, which helps explain why volume flexibility and broad product lines were consistently ranked as relatively low capabilities for Innovators, Servers, and Mass Customizers. We propose that it is appropriate to change the Innovators name to the Specialists. The new name is in better keeping with the notion of this cluster following a focus strategy. It is also more consistent with the observation that Specialists single-mindedly value performance and conformance quality over all other capabilities including price and flexibility. Additionally, given that Specialists place less importance on new product design in absolute terms than do Designers, it seems best to avoid the potential confusion between Designers and Innovators. 7.2. Theory extension While the three strategic groups and their core sets of competitive capabilities remained relatively consistent, the underlying dimensions that manufacturers compete on appear to shift relentlessly. In 1987, Miller and Roth interpreted the underlying strategy dimensions behind the total set of North American competitive capabilities as market differentiation and market scope. By 1994, these underlying dimensions evolved into the cost and after-sales service components of customer value. Just four years later, cost, design flexibility, and broad product lines were important underlying drivers of North American manufacturing strategy. These results indicate that manufacturing strategy is static or dynamic depending on the circumstances. Our longitudinal analysis suggests that while the Caretakers and Specialists were interested in “holding the course,” the Marketeers, cum Designers, seemed equally determined to change directions. The Caretakers’ approach is fundamentally a static manufacturing strategy with consistent anchors around low price, conformance quality, and dependable delivery. Similarly, the Specialists’ strategy is constant with predicted performance and conformance quality attacks combined with a retreat from price competition.
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
The Marketeers appeared to have the only dynamic manufacturing strategy. Porter (1985) noted that while industry structure is relatively stable, it shifts over time, and this leads some companies to change. Not unlike chameleons, Marketeers, and subsequently Designers, seemed willing and able to shift the core elements of their manufacturing strategy to suit the business environment. These shifts, in turn, are what caused the underlying dimensions that all manufacturers compete on to regularly reshuffle. The intriguing question is why the Marketeers and Designers are the ones that keep changing. Was there something fundamentally wrong with their strategy that made it difficult for Marketeers (and likely next the same for the Designers) to permanently follow, or was something else driving their agenda? Differentiated manufacturing strategies like those of the Marketeers and Designers could simply be the hardest to get right, or perhaps given their highly customer-driven nature these strategies are destined to periodically evolve. Conversely, what is it about the Caretakers’ and Specialists’ strategies that predictably leads approximately half of all North American manufacturers and a sizable percentage of international firms to pursue these approaches? The next rounds of research need to concentrate in depth on only one of these strategy types at a time so that we can better understand the motivations behind each strategic group as well as their associated actions. Given their continual evolution, detailed studies of Designers are especially important. Evidence suggests that understanding the drivers of the Designers’ strategy may be a key to predict major trends in our field, since Caretakers and Specialists will presumably continue to emphasize the same capabilities on into the future. It also underlines the importance of going beyond classifying strategies to the issues of strategy implementation, a vital and relatively less researched domain. A second finding from this study is that there are varying concentrations of strategic groups as well as other manufacturing strategy types present around the world. Evidence suggests that while parts of Miller and Roth’s (1994) taxonomy apply globally, there are different ratios of clusters from those typically found in North American. A hypothetical North America sample can be expected to have approximately 50 percent that follow a differentiated strategy like the Marketeers or Designers, while roughly 30 percent are
557
Specialists and the remainder are Caretakers. A global sample will probably contain few, (if any), Specialists and proportionally higher percentages of Designers and Caretakers. There are also at least three other manufacturing strategy types — Idlers, Servers, and Mass Customizers — followed around the world. This has some important implications for researchers studying international manufacturers and suggests that the North American model of manufacturing strategy should not be haphazardly applied to global data. It also raises the question of whether the new strategy types uncovered in this study are endogenous to each region, or if they will eventually diffuse to other parts of the world. As we did with Miller and Roth’s work, our global findings merit replication. While OM has made great strides in operations strategy research during recent years, without a fundamental commitment from other researchers to challenge and validate such results, our field’s future progress will certainly be more limited.
References Aldenderfer, M.S., Blashfield, R.K., 1984. Cluster Analysis. Sage, London. Amundson, S.D., 1998. Relationships between theory-driven empirical research in operations management and other disciplines. Journal of Operations Management 16, 341–359. Arabie, P., Huber, L., 1994. Cluster analysis in marketing research. In: Bagozzi, R.P. (Ed.), Advanced Methods in Marketing Research. Blackwell, Oxford, pp. 160–189. Bacharach, S.B., 1989. Organizational theories: some criteria for evaluation. Academy of Management Review 14 (4), 496–515. Blaug, M., 1992. The Methodology of Economics: Or How Economists Explain. Cambridge University Press, New York. Bozarth, C., McDermott, C., 1998. Configurations in manufacturing strategy: a review and directions for future research. Journal of Operations Management 16, 427–439. Clark, K.B., Fujimoto, T., 1991. Product Development Performance: Strategy, Organization, and Management in the World Auto Industry. Harvard Business School Press, Boston. Dess, G., Newport, S., Rasheed, A.M., 1993. Configuration research in strategic management: key issues and suggestions. Journal of Management 19, 775–795. Dewald, W.G., Thursby, R.G., Anderson, R.G., 1986. Replication in empirical economics: the journal of money, credit, and banking project. American Economic Review 76, 587–603. Eisenhardt, K.M., Tabrizi, B.N., 1995. Accelerating adaptive processes: product innovation in the global computer industry. Administrative Science Quarterly 40 (1), 84–100.
558
M.T. Frohlich, J.R. Dixon / Journal of Operations Management 19 (2001) 541–558
Fine, C.H., 1998. Clock Speed: Wining Industry Control in the Age of Temporary Advantage. Perseus Books, Reading. Flynn, B.B., Sakakibara, S., Schroeder, R.G., Bates, K.A., Flynn, E.J., 1990. Empirical research methods in operations management. Journal of Operations Management 9 (2), 250– 284. Flynn, B.B., Schroeder, R.G., Flynn, E.J., 1999. World class manufacturing: an investigation of Hayes and Wheelwright’s foundation. Journal of Operations Management 17, 249–269. Frohlich, M.T., Westbrook, R., 2001. Arcs of integration: an international study of supply chain strategies. Journal of Operations Management 19 (2), 185–200. Galbraith, C., Schendel, D., 1983. An empirical analysis of strategy types. Strategic Management Journal 4 (2), 153–173. Gilmore, J.H., Pine, J.B., 1997. The four faces of mass customization. Harvard Business Review 75 (1), 91–101. Goh, C., Holsapple, C.W., Johnson, L.E., Tanner, J.R., 1997. Evaluating and classifying POM journals. Journal of Operations Management 15, 123–138. Gronroos, C., 1983, Strategic Management and Marketing in the Service Sector. Marketing Science Institute, Cambridge. Gummesson, E., 1991. Service quality: a holistic view. In Brown, S.W., et al. (Eds.), Service Quality: Multidisciplinary and Multinational Perspectives. Lexington Books, Lexington. Hayes, R.H., Wheelwright, S.C., 1984. Restoring our Competitive Edge: Competing Through Manufacturing. Wiley, New York. Hempel, G.C., 1966. Philosophy of Natural Science. Prentice-Hall, Englewood Cliffs, NJ. Hill, T., 1995. Manufacturing Strategy: Text and Cases. Macmillan, Houndmills. Hubbard, R., Armstrong, J.S., 1994. Replications and extensions in marketing: rarely published but quite contrary. International Journal of Research in Marketing 11, 233–248. Hubbard, R., Vetter, D.E., 1996. An empirical comparison of published replication research in accounting, economics, finance management, and marketing. Journal of Business Research 35, 153–164. Hubbard, R., Vetter, D.E., Little, E.L., 1998. Replication in strategic management: scientific testing for validity, generalizability, and usefulness. Strategic Management Journal 19, 243–254. Judd, C.M., Smith, E.R., Kidder, L.H., 1991. Research Methods in Social Relations, Sixth Edition. Holt, Rinehart & Wilson, Fort Worth, New York, pp. 27. Kane, E.J., 1984. Why journal editors should encourage the replication of applied economic research? Quarterly Journal of Business and Economics 23, 3–8.
Kessler, E.H., Chakrabarti, A.K., 1996. Innovation speed: a conceptual model of context, antecedents, and outcomes. Academy of Management Review 21 (4), 1143–1191. Ketchen, D., Thomas, J.B., Snow, C.C., 1993. Organizational configurations and performance: a comparison of theoretical approaches. Academy of Management Journal 36, 1278–1313. Ketchen, D., Shook, C., 1996. The application of cluster analysis in strategic management research: an analysis and critique. Strategic Management Journal 17 (6), 441–459. Kuhn, T., 1962. The Structure of Scientific Revolutions. University of Chicago Press, Chicago. Lindsay, R.M., 1994. Publication system biases associated with the statistical testing paradigm. Contemporary Accounting Research 11, 33–57. Lindsay, R.M., Ehrenberg, A.S.C., 1993. The design of replicated studies. American Statistician 47, 217–228. Madden, C.S., Easley, R.W., Dunn, M.G., 1995. How journal editors view replication research. Journal of Advertising 24, 77–87. Meyer, A., Tsui, A., Hinings, C., 1993. Introduction: configuration approaches to organizational analysis. Academy of Management Journal 36 (6), 1175–1195. Miller, J.G., Roth, A., 1994. A taxonomy of manufacturing strategies. Management Science 40 (3), 285–304. Pine, J.B., 1993. Mass customizing products and services. Planning Review 21 (4), 6–15. Popper, K., 1978. Conjectures and Refutations (first published in a slightly different form in 1962). Routledge & Kegan Paul, London, pp. 33–39. Porter, M.E., 1985. Competitive Advantage: Creating and Sustaining Superior Performance. The Free Press, New York. Rosenthal, S., 1992. Effective Product Design and Development: How to Cut Lead Time and Increase Customer Satisfaction. Business One Irwin, Homewood. Rosenthal, R., Rosnow, R.L., 1984. Essentials of Behavioral Research: Methods and Data Analysis. McGraw-Hill, New York. Schilling, M.A., Hill, C.W.L., 1998. Managing the new product development process: strategic imperatives. The Academy of Management Executives 12 (3), 67–81. Selltiz, C., Wrightsman, L.S., Cook, S.W., 1976. Research Methods in Social Relations, Third Edition. Holt, Rinehart & Winston, New York, p. 46. Voss, C., Blackmon, K., 1998. Differences in manufacturing strategy decisions between Japanese and Western manufacturing plants: the role of strategic time orientation. Journal of Operations Management 16, 147–158.