Professional Documents
Culture Documents
Strategic Group Analysis in The Construction Industry: Irem Dikmen M. Talat Birgonul and Cenk Budayan
Strategic Group Analysis in The Construction Industry: Irem Dikmen M. Talat Birgonul and Cenk Budayan
Abstract: The aim of strategic group analysis is to determine whether clusters of firms that have a similar strategic position exist within
an industry or not. Findings of strategic group analysis may further be used to investigate the performance implications of strategic group
Downloaded from ascelibrary.org by Universiti Sains Malaysia - Transkrian on 12/14/19. Copyright ASCE. For personal use only; all rights reserved.
membership. The objective of this paper is to identify the possible strategic groups that could exist within the Turkish construction
industry by using a theoretical framework applicable for the construction industry and alternative statistical cluster analysis techniques.
The achieved results pinpoint the existence of three clusters and significant differences between the performances of firms in each cluster.
All of the firms in the strategic group that have the highest average performance utilize a quality differentiation strategy and have the
necessary resources and capabilities that give them the opportunity to differentiate their services from others. Also, they use a systematic
approach and have a collaborative environment for strategic decision making. Findings of strategic group analysis can be used by
professionals to understand the current strategic position of a firm within the competitive environment and formulate strategies to shift to
a better performing cluster.
DOI: 10.1061/共ASCE兲0733-9364共2009兲135:4共288兲
CE Database subject headings: Construction industry; Statistics; Decision making; Competition.
tition is defined as either narrow or broad market approach. A are various researchers that used a broad perspective for defining
questionnaire was prepared in which all items were measured on strategic variables 共Cool and Schendel 1987; Dess and Davis
a 1–5 Likert scale. Performance was measured by using a subjec- 1984; Fiegenbaum and Thomas 1990兲. However, defining a high
tive reporting approach. K-means cluster analysis was used to number of variables from a broad perspective is not without criti-
classify firms according to the identified dimensions 共mode and cisms. Barney and Hoskisson 共1990兲 pointed out the potential for
scope兲. As a result, based on the responses from 107 firms, four introducing noise into the data analysis process as all variables
clusters were identified. Statistically significant performance dif- are usually given equal weight in generating clusters. Thus, if
ferences were found between the clusters. The writers claimed irrelevant factors are included in the analysis, the resulting clus-
that differences in construction companies’ performances can be ters may not reflect the actual group structure. They suggested
partly explained by their choices of mode and scope of competi- that some theoretical framework should be used for choosing stra-
tion. It was found that construction companies that outperform tegic variables and judging the quality of results. However, this
their rivals adopt a hybrid mode, rather than a single mode of theory has yet to be developed.
competition. However no statistically significant finding was Thomas and Venkatraman 共1988兲 suggested utilization of the
found on impact of scope on performance. following strategies for the selection of a relevant conceptual
Claver et al. 共2003兲 studied the linkage of strategic groups and framework:
performance by examining the business strategies of Spanish con- 1. It should match the key basis of competition in the market
tractors. Using the data regarding 88 housing contractors and de- place; and
fining variables that are based on Porter’s generic strategies, the 2. It should bear a strong relationship to some of the common
writers identified four strategic groups. However, the empirical theoretical discussions on strategy types so that a cumulative
findings demonstrate that no significant differences exist between perspective can be developed.
the performances of firms that belong to different groups. Thus, in this research, the findings of previous research studies
about the competitive environment within the Turkish construc-
tion industry are used to identify the strategic dimensions for
Objectives of Research grouping. Moreover, the most common framework used in strate-
gic grouping research, which is Porter’s generic typology, is uti-
The objectives of the research are twofold: lized to identify the variables about the strategic position. Details
1. To carry out strategic group analysis for the Turkish con- regarding the selection of the conceptual model and variables are
struction industry and identify possible clusters having dif- given in the next section.
ferent strategic positions; and
2. To test whether performance differences exist between the
firms in different strategic groups. Selection of Variables
In order to eliminate some of the shortcomings of strategic
group analysis, the following strategies are utilized: Price and Newson 共2003兲 mentioned three dimensions of strategy
1. A conceptual framework, which is thought to be a relevant that can be recognized in every real-life strategic problem situa-
model for the construction industry, is defined. It is based on tion:
the work of Price and Newson 共2003兲 and the strategic vari- 1. Strategy content: product of strategy process;
ables are determined by referring to the previous work of 2. Strategy process: manner in which strategies come about;
Dikmen and Birgonul 共2003兲 on strategic perspective of the and
Turkish contractors; and 3. Strategy context: organizational and environmental circum-
2. Strategic group analysis is conducted using cluster analysis. stances under which strategy process and content are
In order to eliminate some of the problems of using cluster determined.
analysis, alternative statistical clustering techniques are used
and the results are compared to find the best technique to
minimize the risk of dependency of results on the selected Strategy Content
technique. The scope and mode of competition are the products of a strategy
formulation process. An organization distinguishes itself by dif-
ferentiating its offerings in some way. Basically there are two
Determining Conceptual Framework for Strategic generic strategies: price differentiation 共cost leadership as defined
Grouping by Porter兲 and quality differentiation. Companies that utilize price
differentiation strategy charge a lower price for their service/
Thomas and Venkatraman 共1988兲 mentioned that there is an ab- product and this approach seems to be the only way to survive in
sence of a generally accepted scheme for defining strategic di- a market like the construction industry, where the lowest cost
into two categories: price and nonprice factors. If a contractor is sumed that companies may be categorized considering whether
competing on the basis of price, the primary objective is to mini- they carry out systematic strategic planning and whether a col-
mize costs. In the other category, which is quality differentiation, laborative environment exists within the company or not.
a contractor tries to maximize client satisfaction through high
quality products and services, innovations, support services, etc.
Strategy Context
In this category, company resources are organized in such a way
that the client objectives are identified and met successfully. The resource-based view asserts the critical role of resources and
Quality differentiation strategy does not mean that cost issues are capabilities of a firm as the basis for strategic choices and primary
not considered; however, reducing cost is not the primary objec- determinants of profitability. Under the title of strategy context,
tive. tangible 共such as financial resources兲 and intangible resources
Companies should make strategic choices about the markets 共such as experience and relations兲 of companies are considered as
they serve and types of projects they undertake. Thus, the bound- well as the capabilities 共such as managerial and technical capa-
aries of an organization should be determined. Companies may bilities兲. Dikmen and Birgonul’s 共2003兲 work on the strategic per-
have a diversification strategy or a focus strategy which deter- spective of Turkish contractors is referred to for identification of
mines their competitive scope. In this research, two dimensions major strengths and weaknesses. According to the results of the
are identified to group the companies with respect to their com- study mentioned, client relations and financial resources are the
petitive scope: market-level and project-level diversification. major determinants of competitive advantage in the Turkish con-
Market-level diversification is about whether the company oper- struction industry.
ates in the construction market only or sectors related and unre- Finally, based on strategy context, content, and process, 13
lated to construction. Also, the amount of work in the domestic variables are defined. These variables and how they are measured
market and international markets are considered under the head- are depicted in Table 1.
ing of market-level strategy. Project-level diversification has two
dimensions: type of project and type of client. The variety of
projects undertaken by the construction company is questioned as Research Methodology
well as the variety of clients served.
The target population is selected as the members of the Turkish
Strategy Process Contractors Association 共TCA兲 which is an independent and non-
profit professional organization that represents the leading con-
During the strategic group analysis, “realized strategies” of firms struction companies in Turkey. The business volume of its
are evaluated to identify possible similarities or differences be- members encompasses nearly 70% of all domestic and 90% of all
tween firms operating in the same market. A realized strategy has international contracting work done so far by the Turkish con-
two components: deliberate strategy 共plans realized兲 and emer- struction companies. From the beginning of the 1970s up to the
gent strategy 共patterns developed in the absence of intentions兲 present, member companies of TCA have completed over 3,000
共Mintzberg 1987兲. Junnonen 共1998兲 also stressed that realized projects in 63 countries and their business volume abroad has
strategy is composed of both intended and emergent actions reached approximately $65 billion United States. Thus, the stra-
which are sometimes formed, not only formulated. Thus, he ar- tegic group analysis carried out in this research covers only the
gued that the success of strategy process depends on both formal- medium-big size contracting firms in Turkey. Small and local
ization of procedures and the existence of the right climate for firms are excluded from the analysis.
strategy formation. Dikmen and Birgonul 共2004兲 identified the A questionnaire form was designed and submitted to 136
following two mechanisms used during the strategy development members of TCA. In the questionnaire, each representative of the
process: company was requested to give relevant information about the 13
1. Formulation of strategies: a strategic planning system is re- strategic variables. The yearly turnover 共average of the last
quired for formulation of strategies and realization of in- 3 years兲 and age of the company were also questioned. A subjec-
tended strategies. The effectiveness of this mechanism tive reporting approach was used for performance assessment
depends on the existence of a system/structure that supports rather than collecting financial data. Dess and Robinson 共1984兲
strategy formulation; and argued that subjective performance measures are most appropriate
2. Emergence/formation of strategies: strategies, other than in- in examining relative performance within an industry. Each re-
tended ones, may emerge within the organization. Success of spondent evaluated his/her company’s performance considering
these strategies depends on the existence of the right climate the previous 3 year period in terms of profitability, workload, and
and organizational culture that permits idea generation. The other company objectives. The total number of returned question-
effectiveness of this mechanism is mainly about the power naires is 84 and the return rate is 0.62.
culture and management style of the organization rather than “Size” as defined in this study refers to the average turnover
Category 3: industrial
Category 4: others
Type of client 共client type that has the highest percentage in Category 1: government Nominal
total number of projects兲 Category 2: private sector
Strategic planning Category 1: systematic and regular strategic planning Nominal
Category 2: no systematic approach for strategic planning
Strategic decision-making Category 1: democratic and collaborative environment Nominal
Category 2: autocratic approach
Relations with clients Subjective rating using 1–5 scale Ordinal
Human resources Subjective rating using 1–5 scale Ordinal
Managerial capability Subjective rating using 1–5 scale Ordinal
Technical capability Subjective rating using 1–5 scale Ordinal
Financial resources Subjective rating using 1–5 scale Ordinal
Experience Subjective rating using 1-5 scale Ordinal
value recorded for the last 3 years. Thus, a high correlation is ter and two-step cluster analysis. All of these methods have ad-
expected to exist between the performance rating and size, since vantages and disadvantages over each other. For example, the
one of the components of performance rating is “workload.” Cor- K-means clustering method requires that the number of clusters
relations between the strategic variables, size, and age should be are known at the start of the application, whereas in the hierar-
calculated and if age and/or size are found to be strongly corre- chical clustering method, the number of clusters are determined
lated with other variables, the list of variables to be used during after the analysis. On the other hand, in the hierarchical cluster
clustering should be revised. analysis, there is only one pass through the data. Thus, the wrong
partition of data may persist throughout the analysis and lead to
wrong results. K-means cluster analysis can fix an early poor
Cluster Analysis partition problem at later stages of the process by passing over the
data more than once. The most important advantage of the two-
Statistical cluster analysis is used for strategic group analysis of step cluster method is its ability to determine the number of the
the Turkish construction industry. The aim of cluster analysis, as clusters automatically. Also, a large quantity of data can be ana-
defined by Bacher 共2002兲, is finding K clusters so that the objects lyzed efficiently. However, accurate results can be reached only if
of one cluster are similar to each other, whereas objects of differ- all variables are independent, continuous variables have normal
ent clusters are dissimilar. Cluster analysis simply finds the pos- distributions and categorical variables have multinomial distribu-
sible groups without explaining why they exist. Cluster analysis tions. In real life, these conditions are rarely provided 共Norusis
can be used for finding a true typology, model fitting, prediction 2004兲. In this study, as it is impossible to determine the most
based on groups, hypothesis testing, data exploration, hypothesis appropriate clustering technique at the beginning, all three meth-
generation, and data reduction 共Everitt 1974兲. ods are used and results are compared to select the best one.
Before the application of cluster analysis, the scale and vari- There are various algorithms used during hierarchical cluster
ance of the variables and multilinearity among the variables analysis. In the first step, all algorithms of the hierarchical cluster
should be checked. In this research, as no significant differences analysis were used in order to make an initial guess about the
were found between the mean values and standard deviations, number of clusters. The results were presented by a dendrogram
standardization was not applied to the current data set. Also, the and an inverse scree tree was drawn by using an agglomeration
data set was checked for multilinearity, as high correlations be- schedule. Analysis was carried out by using the Manhattan 共city
tween variables can cause flawed results by overweighting one or block兲 distance, as it is the most suitable parameter when the data
more underlying constructs. In this research, Pearson correlations have nominal and ordinal variables 共Bacher 2002兲. Finally, the
between the variables were calculated and it was proved that no alternative number of clusters was determined as two or three. In
multilinearity effect exists in the current data set. Also, size and the next step, two-step cluster analysis was conducted where the
age were not strongly correlated with other variables. The corre- number of clusters was determined automatically by using
lation matrix is presented in Table 2. Schwarz’s Bayesian criterion 共BIC兲 and Akaike’s information
The tool used for cluster analysis is SPSS 13.0, which offers criterion 共AIC兲. The procedure was carried out using these two
three methods for clustering: Hierarchical cluster, K-means clus- criteria separately and finally, the number of clusters was deter-
9. Human resources 0.30a 0.29a −0.02 0.20 −0.03 0.46b 0.38b 0.64b 1.00 — — — — — — —
10. Managerial capability 0.27 0.17 −0.05 0.24 0.03 0.55b 0.31a 0.53b 0.58b 1.00 — — — — — —
11. Technical capability 0.27 0.37b 0.02 0.07 −0.01 0.46b 0.46b 0.56b 0.65b 0.64b 1.00 — — — — —
12. Financial resources 0.22 0.15 0.09 −0.02 −0.05 0.48b 0.44b 0.60b 0.61b 0.54b 0.61b 1.00 — — — —
13. Experience 0.29a 0.30a 0.03 0.12 −0.03 0.59b 0.52b 0.61b 0.65b 0.55b 0.64b 0.50b 1.00 — — —
14. Differentiation strategy 0.18 0.02 0.05 0.18 −0.08 0.45b 0.42b 0.62b 0.52b 0.47b 0.50b 0.48b 0.44b 1.00 — —
15. Diversification strategy 0.40b 0.18 −0.03 −0.10 0.16 0.04 0.11 0.02 0.13 0.27 0.26 0.15 0.18 −0.01 1.00 —
16. Performance 0.30a 0.27 0.01 0.16 −0.00 0.62b 0.48b 0.75b 0.75b 0.65b 0.76b 0.75b 0.74b 0.65b 0.18 1.00
a
Correlation is significant at the 0.01 level 共two-tailed兲.
b
Correlation is significant at the 0.001 level 共two-tailed兲.
mined as three. Finally, by referring to the findings of previous whether small modifications in the data set change the outputs of
methods, K-means cluster analysis was carried out for two and the method or not. Nine firms 共around 10% of the population兲
three clusters. The distance between the centers of the clusters were randomly drawn from the data set and the Rand indices were
was determined to be 3.843 for the two-clusters solution. For the calculated for each method. The Rand indices of two-step cluster
three-clusters solution, the distances were found to be 2.147, analysis with Schwarz’s Bayesian criterion, average linkage,
5.491, and 3.475 between the first and second, first and third and K-means, and Ward’s methods were the highest among all. The
the second and third clusters, respectively. The number of clusters number of members in each cluster was also checked for each
was chosen as three, a since three-cluster solution may provide method. The methods which provide clusters that have only one
some significant information that could be lost in the two-cluster or two members should not be used 共Bacher 2002兲. In this data
solution. set, the average linkage method yielded a cluster with two mem-
After deciding on the number of clusters, some tests must be bers, the single linkage method resulted in two clusters with one
carried out to check stability, relative size of the clusters, and member, the centroid linkage method lead to one cluster with two
external validity. The Rand index, which measures the proportion members, and the solution of the median linkage method sug-
of consistent allocations, should be calculated in order to check gested one cluster with one member and one cluster with two
stability. If the Rand index is equal to one, it means a perfect fit. members. These methods were excluded and the external validity
Values over 0.7 are considered as sufficient for the stability of the test was performed for the remaining methods. For the external
cluster analysis 共Bacher 2002兲. Table 3 gives the Rand indices for validity, the performance rating was used as the external variable
different methods. All of the Rand indices of Ward’s method, and analysis of variance 共ANOVA兲 analysis whose results are
complete linkage, and two-step cluster analysis were found to be shown in Table 4 was performed for each method.
higher than 0.7. The stability of the analysis was also checked by According to ANOVA results, it can be concluded that perfor-
drawing some random samples from the population and testing mance differences exist between clusters in all of the methods. By
comparing the F values, it can be noted that the clusters obtained When compared to the other clusters, the ratings regarding the
Downloaded from ascelibrary.org by Universiti Sains Malaysia - Transkrian on 12/14/19. Copyright ASCE. For personal use only; all rights reserved.
by using the K-means method are the most dissimilar clusters. In level of experience and client relations are significantly higher.
addition to this, the K-means method shows the highest stability The values of other variables regarding the strategy context,
for the data set. Therefore, the clusters found by this method can which are financial resources, technical capability, human re-
be used for strategic grouping. Table 5 shows the clusters sug- sources, and managerial capability, are also very high 共in the
gested by this method. ANOVA results are depicted in Table 6. range of 4–5兲. All of the firms in this group utilize a nonprice
strategy, meaning that they differentiate themselves from others in
terms of quality, innovation, etc. Similarly, the majority of the
Discussion of Research Findings firms are diversified into sectors unrelated to construction. The
average performance rating is the highest among the three groups.
ANOVA results depicted in Table 6 demonstrate the variables that This may result from the exceptional level of resources and capa-
significantly contribute to the formation of the clusters. The top bilities which give them the opportunity to differentiate their ser-
three factors that show the highest contribution to the separation vices from others. Also, unrelated diversification may result in
of the clusters are: experience, relations with clients, and human financial synergy leading to the highest financial resources score
resources. About the strategy process, both types of strategic plan- among the three groups.
ning and decision-making processes significantly contribute to the
differences in clusters. Similarly, differentiation and diversifica- Strategic Group 2. This group comprises 36 companies with
tion strategies have significant contributions. The contributions of an average yearly turnover value of $47.17 million United States
factors such as internationalization ratio, type of projects, and $/year. This value is lower than the average of Strategic Group 1.
clients are found to be insignificant 共if 5% significance level is All of the strategy context variables have values within the range
considered兲. One can argue that strategy context is the major de- of 3–4, showing that their competitiveness is lower in all dimen-
terminant of performance differences in different strategic groups. sions when compared with the firms that belong to Strategic
In light of the statistical findings, the following three strategic Group 1. Also, the percentage of firms having systematic strategic
groups are found to exist within the Turkish construction industry. planning and democratic decision-making process is significantly
lower. This also holds true for the strategy content. It may be
Strategic Group 1. The number of firms in this group is 34. argued that although the majority of the firms in this group have
This group comprises the biggest firms 共average turnover is nonprice differentiation strategy, the level of resources and com-
around $108 million $/year兲. The majority of the firms in this petencies that backup this strategy is not as high as in Strategic
group have a systematic strategic planning process and strategies Group 1. Also, it is evident that the majority of the firms in this
are formulated in a democratic decision-making environment. group utilize focus strategy rather than diversification. The per-
same impact on all firms and its implications are reflected in the formance and information technology utilization in the construction
performance ratings in the same manner. Also, the research has industry.” J. Constr. Eng. Manage., 132共5兲, 499–507.
some limitations due the measurement method used. Single items Everitt, B. 共1974兲. Cluster analysis, Wiley, New York.
are used to measure variables rather than constructs composed of Fiegenbaum, A., and Thomas, H. 共1990兲. “Strategic groups and perfor-
a series of questions. Moreover, performance rating is subjective mance: The U. S. insurance industry, 1970–84.” Strategic Manage. J.,
and reflects the personal judgment of only one respondent. 11共3兲, 197–215.
If a rational theoretical framework is used together with an Fleisher, C., and Bensoussan, B. 共2000兲. “A FAROUT way to manage CI
appropriate clustering method, strategic group analysis may pro- analysis.” Competitive Intell. Mag., 3共2兲, 37–40.
Frazier, G. L., and Howell, R. D. 共1983兲. “Business definition and per-
vide significant information for analyzing intraindustry heteroge-
formance.” J. Retailing, 47共2兲, 56–67.
neity. In order to increase the reliability of results, the cognitive
Grant, R. M. 共1995兲. Contemporary strategy analysis: Concepts, tech-
approach can be used together with statistical tools and their re-
niques, applications, 2nd Ed., Blackwell Publishing, Oxford, U.K.
sults may be compared. Another topic for further study can be
Hampson, K., and Tatum, C. B. 共1997兲. “Technology strategy and com-
conducting strategic group analysis over time and monitoring per- petitive performance in bridge construction.” J. Constr. Eng. Manage.,
formance implications with changing strategic positions and com- 123共2兲, 153–161.
petitive rules. Dynamic strategic group analysis can be carried out Hasegawa, F. 共1988兲. Built by Japan: Competitive strategies of the Japa-
so that one can question the relevance of strategic group structure, nese construction industry, Wiley, New York.
identify causes of performance differences between groups over Hatten, K., and Hatten, M. L. 共1987兲. “Strategic groups, asymmetrical
time, and finally determine whether mobility barriers exist or not. mobility barriers and contestability.” Strategic Manage. J., 8共4兲, 329–
Finally, this study statistically proves that different strategic 342.
groups may exist within the construction industry and strategic Hodgkinson, G. P. 共1997兲. “The cognitive analysis of competitive struc-
group membership may be a determinant of performance. How- tures: A review and critique.” Hum. Relat., 50共6兲, 625–654.
ever, results are country specific and cannot be generalized. Com- Hunt, M. 共1972兲. “Competition in major home appliance industry 1960–
parative studies can be conducted by using the strategic 1970.” Ph.D. dissertation, Harvard Univ., Cambridge, Mass.
dimensions and analysis method proposed in this paper. Jennings, M., and Betts, M. 共1996兲. “Competitive strategies for quantity
surveying practices: The importance of information technology.”
Eng., Constr., Archit. Manage., 3共3兲, 163–186.
Junnonen, J. M. 共1998兲. “Strategy formation in construction firms.” Eng.,
References Constr., Archit. Manage., 5共2兲, 107–114.
Kale, S., and Arditi, D. 共2002兲. “Competitive positioning in United States
Akintoye, A., and Skitmore, M. 共1991兲. “Profitability of UK construction construction industry.” J. Constr. Eng. Manage., 128共3兲, 238–247.
contractors.” Constr. Manage. Econom., 9共4兲, 311–325. Kale, S., and Arditi, D. 共2003兲. “Differentiation, conformity, and con-
Amel, D. F., and Rhoades, S. A. 共1988兲. “Strategic groups in banking.” struction firm performance.” J. Manage. Eng., 19共2兲, 52–59.
Rev. Econ. Stat., 70共4兲, 685–689. Ketchen, D. J., and Shook, C. L. 共1996兲. “The application of cluster
Bacher, J. 共2002兲. “Cluster analysis.” Script.pdf, 具http://www.soziologie. analysis in strategic management research: An analysis and critique.”
wiso.uni-erlangen.de/koeln/script/script.pdf典 共Jan. 8, 2007兲. Strategic Manage. J., 17共6兲, 441–458.
Barney, J. B., and Hoskisson, R. E. 共1990兲. “Strategic groups: Untested Leask, G. 共2004兲. “Strategic groups and the resource based view: Natural
assertions and research proposals.” Manage. Dec. Econ., 11共3兲, 187– complements enhancing our understanding of the competitive pro-
198. cess.” Working paper, Ashton Univ., Birmingham, U.K.
Betts, M., and Ofori, G. 共1992兲. “Strategic planning for competitive ad- Lewis, P., and Thomas, H. 共1990兲. “The linkage between strategy, strate-
vantage in construction.” Constr. Manage. Econom., 10共6兲, 511–532. gic groups, and performance in the U.K. retail grocery industry.” Stra-
Caves, R., and Porter, M. E. 共1977兲. “From entry barriers to mobility tegic Manage. J., 11共5兲, 385–397.
barriers: Conjectural decisions and contrived deterrence to new com- Mascarenhas, B. 共1989兲. “Strategic group dynamics.” Acad. Manage J.,
petition.” Q. J. Econ., 91共2兲, 241–262. 32共2兲, 333–352.
Chinowsky, P. S., and Meredith, J. E. 共2000兲. “Strategic management in Mascarenhas, B., and Aaker, D. A. 共1989兲. “Mobility barriers and strate-
construction.” J. Constr. Eng. Manage., 126共1兲, 1–9. gic groups.” Strategic Manage. J., 10共5兲, 475–485.
Claver, E., Molina, J. F., and Tari, J. J. 共2003兲. “Strategic groups and firm McGee, J., and Thomas, H. 共1986兲. “Strategic groups: Theory, research,
performance: The case of Spanish house-building firms.” Constr. and taxonomy.” Strategic Manage. J., 7共2兲, 141–160.
Manage. Econom., 21共4兲, 369–377. McNamara, G., Deephouse, D. L., and Luce, R. A. 共2003兲. “Competitive
Cool, K. O., and Schendel, D. 共1987兲. “Strategic group formation and positioning within and across a strategic group structure: The perfor-
performance: The case of the U.S. pharmaceutical industry.” Manage. mance of core, secondary, and solitary firms.” Strategic Manage. J.,
Sci., 33共9兲, 1102–1124. 24共2兲, 161–181.
Dess, G. G., and Davis, P. 共1984兲. “Porter’s 共1980兲 generic strategies as Mintzberg, H. 共1987兲. “Five Ps for strategy.” California Manage. Rev.,
determinants of strategic group membership and organizational per- 30共1兲, 11–24.
formance.” Acad. Manage J., 27共3兲, 467–488. Norusis, M. 共2004兲. SPSS 13.0 statistical procedures companion,
Dess, G. G., and Robinson, R. B. 共1984兲. “Measuring organizational per- Prentice-Hall, Upper Saddle-River, N.J.
ation of paradoxes, processes, and associated concepts as applied to J. Constr. Eng. Manage., 122共2兲, 133–140.
construction.” J. Manage. Eng., 19共4兲, 183–192. Wiggins, R. R., and Ruefli, T. W. 共1995兲. “Necessary conditions for the
Reger, R. K., and Huff, A. S. 共1993兲. “Strategic groups: A cognitive predictive validity of strategic groups: Analysis without reliance on
perspective.” Strategic Manage. J., 14共2兲, 103–124. clustering techniques.” Acad. Manage J., 38共6兲, 1635–1656.