MULTIDIMENSIONALITY EVALUATION OF SUPPLY CHAIN MANAGEMENT INTEGRATION

Integration is constantly reported in the literature as an essential feature of SCM. However, it is as difficult to define as to operationalize it, resulting in a lack of information on how to increase the level of integration among members. Characteristics such as trust, sharing of information, partnership, cooperation, collaboration and coordination are constantly associated to the definition of SCM. However, further studies are needed relating the contribution of each characteristic separately, neglecting the multidimensional aspect of SCM. Thus, this research aims to evaluate the dimensionality of SCI, to check alignment with the features identified in the literature using multivariate statistical analysis. The methodology used was based on a questionnaire to assess the level of companies integration with suppliers of its supply chain, obtaining 205 answers. Full-information item factor analysis and principal component analysis on the correlation matrix tetrachoric were used for the instrument dimensionality analysis.

However, further studies are needed relating the contribution of each characteristic separately, neglecting the multidimensional aspect of SCM. Thus, this research aims to evaluate the dimensionality of SCI, to check alignment with the features identified in the literature using multivariate statistical analysis. The methodology used was based on a questionnaire to assess the level of companies integration with suppliers of its supply chain, obtaining 205 answers. Full-information item factor analysis and principal component analysis on the correlation matrix tetrachoric were used for the instrument dimensionality analysis.

INTRODUCTION
Integration is constantly reported in the literature as an essential feature of the Supply Chain Management (SCM) (NÄSLUND; HULTHEN, 2012; PEARCY; GIUNIPERO, 2008;RICHEY JUNIOR et al., 2009). However, it is as difficult to define integration as operationalize it, resulting in a lack of information on how to increase the level of SCI (Supply Chain Integration) (NÄSLUND; HULTHEN 2012).
The literature defines SCI as the alignment of internal and external flows of a supply chain through collaboration and coordination among members, seeking effective and efficient flow of goods, services, information and financial, to generate value for the end customer (JÜTTNER et al., 2007;FABBE-COSTES;JAHRE, 2008;FLYNN et al.;, THUN, 2010NÄSLUND;HULTHEN, 2012).
However, further studies are needed aiming to relate the contribution of each
Part of the difficulty in addressing individual aspects of each characteristic may be ascribed to a lack of clarity in the definitions found in the literature. They generally occur as synonyms, alternating the terms that refer to the same behaviors. This collaboration is addressed by Aryee et al. (2008) as a synonym of SCM. Danese (2013)  The understanding of SCM as a multidimensional concept is still scarce in the literature. However, it is of utmost importance so that the integration be increased simultaneously according to different practices (DANESE; BORTOLOTTI, 2014).
Trust consists in believing in the partner's integrity (MORGAN;HUNT, 1994).
Communication, or sharing of information, is essential to a close understanding and cooperation with suppliers and clientes, allowing a broad identification of the clients' requirements (ZHAO et al., 2011). Partnership demands from companies a structural change in how they relate to each other (MALONI; BENTON, 1997).
Coordination encompasses all efforts in aligning decisions to achieve the global objectives of the system (CAO et al., 2008). The identification of how each characteristic is related to an increase in the integration level between the members of the supply chain will allow managers to better plan their actions, working each characteristic according to its specific traits.
This study aims to evaluate the multidimensionality of SCM to allow a better planning of future actions. Thus, a set of SCM indicators was established and applied, as a questionnaire, to a sample of 205 company-supplier relationships. The answers were analyzed based on multivariate statistical analysis, associating indicators with predominant dimensions. http://www.ijmp.jor.br v. 9, n. 1, January -March 2018ISSN: 2236 The SCI has been seen as a way to develop competitive advantage from the management of the relationships, because it reduces the response time to market changes (SEZEN, 2008;KIM, 2009), enabling cost savings by streamlining processes and eliminating redundancies ROSENZWEIG et al., 2003).

INDEPENDENT JOURNAL OF MANAGEMENT & PRODUCTION (IJM&P)
However, it is still common to see companies ignoring the value of investing in internal and external relationships to create competitive advantage. Hence, Souza et al. (2004) believe that knowledge of practices that add value lacks to companies, therefore, they seek immediate and one-off solutions, leading to under-utilization of their potential.
The benefits of integration are often translated into aspects valued by customers, such as product quality, delivery reliability, process flexibility and cost leadership (ROSENZWEIG et al., 2003). However, since the market has customers increasingly demanding and companies, it can no longer overcome on its own. In order to achieve integration, the members of the chain should see each other as partners, working together in the development of strategic planning, demand forecasts and setting of targets (KIM; LEE, 2010). However, to achieve SCI, some characteristics must be identified in the relationship between members, they are: trust, information sharing, partnership, cooperation, collaboration and coordination (ARANTES et al. 2014).
This concept applied to the management of relationships between businesses is the basis for the construction of SCI, since, , joint work involves interdependence and therefore it is necessary to depend on the other to achieve their goals (MAYER et al. 1995).
Thus, confidence in the integrity of the partners leads companies to cooperate, since the current relationships provide resources, opportunities and superior benefits to its competitors, with exchange standard at the same level, strategic information sharing and lack of opportunist behavior (MORGAN; HUNT 1994).
The performance of trust can be classified by the fulfillment of high performance promises , by the existence of standard procedures
It is defined as the accomplishment of actions that benefit the other, hoping that this is not used for their own benefit, but as best for both (ARANTES et al. 2014). Some steps to building a relationship based on trust are: consider the relationship value, stipulate operational tasks and decision rights for each party, create effective contracts and design effective solutions to conflicts (CHOPRA; MEINDL, 2003).
The wants and needs of consumers are constantly changing and companies need to adapt to in order to remain competitive. The best way to do it is by establishing a close relationship between customers and suppliers through accurate flow of information demand, which will reduce the time spent in production planning, decrease inventory and make more business sensitive to customer needs (FLYNN et al., 2010;CHOPRA;MEINDL, 2003).
Information sharing can happen at various levels, ranging from complete absence to full information sharing, being an essential factor in reducing the bullwhip effect, which is the distortion of demand information the further the member is from information in the chain (SAHIN; ROBINSON, 2002).
In order to increase efficiency in sharing information, members of the chain can adopt compatible information systems, facilitating problem solving and strategic decision making in a collaborative way (ZHOU; BENTON 2007, HA et al. 2011).
Thus, information sharing in the supply chain, aiming to increase the level of integration among members, is the exchange of strategic information that favor the creation of competitive advantage for the chain as a whole (ARANTES et al. 2014). The partnership requires, from companies, a structural change in the way of relating, encouraging mutual planning and solving problems together (MALONI; BENTON, 1997). It also requires members to be open to make adjustments in the relationship (CHEN; WU, 2010;MOTWANI et al., 1998), that they share capacity, risks, losses and gains (Vieira 2006), have availability and flexibility to adapt to changes (CHEN; WU 2010), in addition to work with fewer suppliers so that it is possible to develop a close relationship with each of its partners (MALONI; BENTON, 1997;MAHESHWARI ET AL. 2006, CHRISTOPHER;JÜTTNER 2000, LAMBERT;COOPER 2000).
Cooperating is "a mean to achieve a certain goal and not an end in itself". It refers to a way of working together that generates benefits for all parts involved in the process. Cooperation occurs when two or more entities come together to obtain benefits that can't be achieved individually, sharing resources such as confidential information, infrastructure, defining standards that improve the interoperability of their systems, optimizing the tactical and operational planning of logistics activities (AUDY et al. 2010).
Collaboration can be defined as the ability to work beyond organizational boundaries to build higher value-added and increase the ability to meet customer needs. It's not only about managing transactions, but developing and implementing new approaches to problem solving considering trust as a basic principle.
Collaboration can be governed by contracts or informally (FAWCETT et al., 2008).
Supply chain processes coordination consists on the intensity in which a company can structure its operational processes, the sharing of resources, rewards and risks in the organizations, in order to become more competitive in the market in which it operates (YEUNG et al., 2009;SIMATUPANG et al., 2002). Coordination creates understanding among members, shapes human behavior and improves competitiveness (MISHRA; SHARMA, 2015). In other words, coordination is to organize the activities of two or more groups so that they are aware of one another's activities and can work together efficiently (SINGH, 2011).
Consequently, it can be concluded that, as part of SCI, cooperation and collaboration can be seen as carrying out activities together, seeking greater gains
From the raised features, SCI indicators that are related to each finding disagreement in the literature about the relationship of the indicators with the characteristics of SCI were identified, ( Table 1). Part of this difference is justified because there is a dependency relationship between some characteristics, since trust is cited as information sharing prerequisite (LAAKSONEN et al. 2009;CHENG et al., 2010), partnership (WEI et al., 2012;LAAKSONEN et al. 2009), collaboration (SPEKMAN et al., 1998;FAWCETT et al., 2008) and cooperation (MORGAN;HUNT, 1994); as well as information sharing is cited as a prerequisite for partnership (SPEKMAN et al.;DU et al., 2012), collaboration (WIENGARTEN et al., 2010;HA et al., 2011) and cooperation (MORGAN;HUNT, 1994;WEI et al., 2012;JÜTTNER et al. 2007).
Trying to find a consensus among these opinions, this research brings together the SCI indicators using full-information item factor analysis, and lists the factors obtained with the features identified in the literature. This allows the understanding of what the characteristics that have the greatest influence on each share of integration between company and suppliers are.

METHODS
The selection of the research method is one of the key decisions to be taken in its construction process, since the collection of data requires a careful and systematic planning when it comes to scientific research (LUDKE; ANDRÉ, 1986).
Thus, for the identification of SCI characteristics review of the literature method proposed by Ensslin et al. (2010) was used, named ProKnow-C Knowledge Development Process-Constructivist.  Audy et al. (2010) This method makes possible the information systematization in a simple way through the bibliometric techniques and the researcher judgment as to what is relevant, following the steps as shown in Figure 1.  Therefore, considering that there is still a high set of articles, the group representing 80% of the total number of citations was determined as the cutoff point.

INDEPENDENT JOURNAL OF MANAGEMENT & PRODUCTION (IJM&P)
The 473 articles total 6,992 citations; considering the cutoff point; the texts that are part of the most cited set are the ones that add up to 80% of this amount. Hence, the articles that make up this group, in this research, are those who received 14 citations or more; which resulted in 126 titles.
From the confirmation of scientific articles recognition by the number of citations, their abstracts are read and then it is defined who will be part of the final

INDEPENDENT JOURNAL OF MANAGEMENT & PRODUCTION (IJM&P)
http://www.ijmp.jor.br v. 9, n. 1, January -March 2018 ISSN: 2236-269X DOI: 10.14807/ijmp.v9i1.664 repository. Among the 126 most cited; 64 abstracts were considered aligned to the research topic, but five of them had the complete text available on the accessed basis, remaining than 59 articles with relevant volume of citations; with abstract aligned with the theme and full text available.
The 347 articles that have been cited a few times or that have not been cited are not discarded; they are divided into two groups and analyzed separately: more recent articles (published after 2011) and older articles. Whereas the newest texts had not enough time to receive a significant number of citations, it is necessary to read their abstracts to identify those which would make the final portfolio of articles, given its alignment with the theme. From 347 articles, 182 were published in the last two years, and 109 were discarded for not presenting abstract alignment with the subject of research, 20 were discarded for not having the full text is available in basis and 53 were selected to compose the articles portfolio by abstract alignment. After reading the complete texts; we came up to a final set of 46 articles that led to the identification of the SCI characteristics cited. From that, the search process shown in Figure 1 was repeated for each one of the identified features as shown in Table 2, always combined with the term "supply chain" located in the abstract.

Keyword
Step selection

Sample and data collection
From SCI characteristics, a questionnaire was designed to assess the level of companies' integration with suppliers of their supply chain. This instrument consisted of 21 items, presented in Table 3, with response options: yes, no or not applicable.
Table3: Questionnaire to assess the level of SCI

N Item
I1 Is the quality of products consistent with the expected?
I2 Do good personal relationships help maintain this relationship?
I3 Is there a standard procedure to replace damaged goods at delivery?
I4 Is there a standard procedure to exchange products under warranty? One of the issues to be taken into account in the evaluation of supply chains is the varying need for managing and the integration level between relationships. It depends on the relevance of each member to the chain (LAMBERT; COOPER 2000). In order to evaluate the integration of the supply chain, an instrument must be used alowing observing the relationships individually (CASTRO et al., 2015; VAN DER VAART; VAN DONK, 2008). Thus, the questionnaire was applied analyzing the company-supplier relationships.

INDEPENDENT JOURNAL OF MANAGEMENT & PRODUCTION (IJM&P)
The answers were collected by a structured interview with managers of 41 companies. The question were about the integration with some of the companies' suppliers using the questionnaire shown in Table 3. The application of the questionnaire using interviews obtains a higher accuracy of answers, minimizing mistakes in the interpretation of questions and ensuring a greater return rate, different from questionnaires sent by e-mail (Bêrni et al., 2012). Each company answered on average 5 questionnaires, resulting in a sample of 205 relationships.
The number of questionnaires answered by each company varies according to the number of suppliers and the availability of each respondent.
The questionnaires application was concentrated in the city of Joao Pessoa, State of Paraiba, Brazil, 77% of the responses were collected in this city; 20% in nearby cities and the remaining 3% in other states. The respondents sample comprises basically micro and small businesses, which represent over 25% of the http://www.ijmp.jor.br v. 9, n. 1, January -March 2018ISSN: 2236 Brazilian GDP. They were selected at random according to the availability of their managers. Most of the sample is concentrated in the sectors of trade, food, construction and manufacturing industry with 44%, 20%, 17% and 10% share respectively.

Data analysis
SCI is not well defined in the literature; it is important performing tests proving the relationship of the proposed features in the explanation of integration. In this sense, multivariate statistics allows reducing and grouping the data, relating the variables, even in the absence of a structured theoretical model (BAKKE et al., 2008).
An important tool of multivariate analysis is the factor analysis. It consists of a class of multivariate statistical methods, in order to define an underlying structure in a data matrix (HAIR Jr et al., 2006). Besides being an objective technique to identify common variability dimensions in a set of variables (CORRAR et al., 2006).

It can be applied when there is a large number of correlated variables in order
to identify a lower number of new alternative variables, which somehow, summarize key information of the original variables in factors or latent variables.
This data summarization enables better information management generating more significant variables, easy to work with (COSTA, 2007). In summarizing the data in this way, the factor analysis captures the latent dimensions that represent the data set on fewer concepts than the original individual variables (COOPER; SCHINDLER, 2003). It is therefore an interdependence technique that evaluates all variables simultaneously, each one connected with the other, using the concept of statistical variable.
Factor analysis is often used when a determination of dependent and independent variables is not previously identified (COOPER; SCHINDLER, 2003), since this method makes it possible to relate the variables based on their correlation, forming groups of highly interrelated variables (BAKKE et al., 2008). These groups represent dimensions that, together, can explain the integration (HAIR JR et al., 2006). Thus, the variables are combined in accordance with the latent integration feature that they represent.
Factor analysis assumes that the variables are continuous and it is based on the correlation matrix of those variables. For dichotomous variables, it is possible to http://www.ijmp.jor.br v. 9, n. 1, January -March 2018 ISSN: 2236-269X DOI: 10.14807/ijmp.v9i1.664 adjust the factor analysis by adopting the matrix of tetrachoric correlation (BARTHOLOMEW; KNOTT, 1999). A more recent approach is called full-information item factor analysis, that uses the respondents answers standards adjusting the response theory models to the multidimensional item (WIRTH; EDWARDS, 2007;BOCK;GIBBONS, 1988).

RESULTS AND DISCUSSION
The different latent features that compose the SCI can be seen through the graph of eigenvalues of tetrachoric matrix correlation, consisting of the variance explained by each factor shown in Figure 2. This graph is the result of principal component analysis, and factor analysis, it reduces the number of the original set variables in main components, generating new coordinates, simpler to be analyzed.
The analysis of this graph indicates that the set of items used to assess the SCI can be grouped in approximately 4 factors.

Figure 2 -Eigen values of tetrachoric matrix
The percentage of variance explained by the full-information item factor analysis model (Table 4) indicates that 72% of the model change can be explained with 4 factors, and 79% if 5 factors are considered. By this criterion, it is reasonable to consider the number of factors that exceeds 70% of variance explained by the model, and, according to Reckase (2009), the aim of factor analysis is to find the

INDEPENDENT JOURNAL OF MANAGEMENT & PRODUCTION (IJM&P)
http://www.ijmp.jor.br v. 9, n. 1, January -March 2018 ISSN: 2236-269X DOI: 10.14807/ijmp.v9i1.664 smallest number of factors that explain the latent variable. However, it is important to note the information gain by adding another factor, in addition to observing the theoretical coherence of the proposed pool.
Table4: Percentage of variance explained by the model A way to verify the information gain with an increase in the number of dimensions is to observe the Akaike (AIC) and Bayesian (BIC) information criterion shown in Table 5 http://www.ijmp.jor.br v. 9, n. 1, January -March 2018ISSN: 2236 collaboration and coordination. The results reinforce the idea according to which integration is a multidimensional concept (VALLET-BELLMUNT; RIVERA-TORRES, 2013) and that such aspect must be taken into account to improve the integration level in their different practices (DANESE; BORTOLOTTI, 2014).
The feature that should be seen as the basis for constructing the integration in a supply chain is trust among members. Feature that involves maintaining the quality standard of the product supplied , the existence of standard procedure to replace damaged goods on delivery or within warranty (CHOPRA; MEINDL, 2003;FAWCETT ET AL., 2012) and good personal relations between the parties (CAI et al., 2010), which will favor the development of the business relationships. being tolerant to any delay in payment and providing a quality product that meets the needs of the end customer.
Although information sharing is considered as a basic feature of SCI, certain information is shared only when companies already have a high level of integration also influencing on how this information is shared among members of the supply chain. These aspects of SCI can be defined as coordination with companies making decisions and solving problems together, sharing cost information with supplier and facilitating access to their information using information systems compatible (CAO et al., 2008;SINGH, 2011;ZHAO et al., 2011). Table 6 deserve attention, as the factor load in item 8, which seeks to know if the company receives updated information from the supplier of the inventory involved. This item presents problems of interpretation, which generated a lower result than expected. On the other hand, the last two items on the coordination had factor loadings higher than 1, which according to Jöreskog (1999), may occur when the factors are related, indicating that the greater the load factor, the greater the ratio between factor and items.

Some of the values shown in
This variables reduction, provided by factor analysis, allows us to better understand how to increase the level of integration among the members of a supply chain, since we can work with less complex variables. Based on it, we can individually study each of the identified features, combining their individual ways to increase the level of SCI.
However, it is noteworthy that some actions are more complex than others.
This implies that the SCI must be installed in a gradual way and its evolution depends on how relationships with partners in the supply chain are conducted. As trust among members increases, and more joint actions are developed, the performance of the supply chain increases, generating higher earnings for the most integrated members.

CONCLUSION
The SCI is a very broad topic, covering different characteristics and this has led the authors to have different definitions that vary according to the approach used in their analysis, limiting planning the construction of an integrated supply chain. The evaluation of SCM muldimensionality allowed separating different practices according to a predominant characteristic. To acknowledge one of these
SCM is a multidimensional concept. To find ways to improve integration in supply chains, it must be seen as such. This is an aspect little studied by the existing literature on this topic. This study is a small step in that direction. However, much must still be discussed on this topic.
One limitation found in studies on companies is representative samples because of the lack of availability to participate in academic research (BORTOLOTTI, 2010). Thus, assessing the dimensionality of the instrument in another sample is needed to determine the generalizability of the results (Immekus and Imbrie 2014).
As further research, we suggest the broadening of each characteristic related to SCM. In addition to deciding whether an instrument for measuring the SCI level is expected to generate a total score for all items or if scores are generated for each dimension.