Supply chain integration scales validation and benchmark values

Purpose: The clarification of the constructs of the supply chain integration (clients, suppliers, external and internal), the creation of a measurement instrument based on a list of items taken from earlier papers, the validation of these scales and a preliminary benchmark to interpret the scales by percentiles based on a set of control variables (size of the plant, country, sector and degree of vertical integration). Design/methodology/approach: Our empirical analysis is based on the HPM project database (20052007 timeframe). The international sample is made up of 266 plants across ten countries: Austria, Canada, Finland, Germany, Italy, Japan, Korea, Spain, Sweden and the USA. In each country. We analized the descriptive statistics, internal consistency testing to purify the items (inter-item correlations, Cronbach’s alpha, squared multiple correlation, corrected item-total correlation), exploratory factor analysis, and finally, a confirmatory factor analysis to check the convergent and discriminant validity of the scales. The analyses will be done with the SPSS and EQS programme using the maximum likelihood parameter estimation method. Findings: The four proposed scales show excellent psychometric properties. Research limitations/implications: with a clearer and more concise designation of the supply chain integration measurement scales more reliable and accurate data could be taken to analyse the relations between these constructs with other variables of interest to the academic l fields.


Introduction
The concept of supply chain integration is of great interest for academics working in operational management (Zhao, Huo, Selen & Yeung, 2011;Flynn, Huo & Zhao, 2010). One of the main reasons is that it greatly influences the competitive advantage of companies Chang, Ik-Whan & Dennis, 2007;Alfalla-Luque, Medina-Lopez & Schrage, 2012). But it is also a concept whose definition and whose operationalization are still up for debate. There is no consensus as to which components to include, nor how to measure them (Roth, Schroeder, Huang & Kristal, 2008;Zhu, Sarkis & Lai, 2008;Li, Rao, Ragu-Nathan & Ragu-Nathan, 2005;, Alfalla-Luque, Medina-Lopez & Dey, 2012. In fact, in research carried out so far, it is common to be confronted with a variety of proposals and this means that demonstrating the effects of supply chain integration on the performance of companies is inconclusive giving contradictory results (Zhao et al., 2011;Chang et al., 2007;. According to recent research, supply chain integration is comprised of two primary dimensions: internal integration and external integration. External integration can then be further subdivided: integration with clients and integration with suppliers (Chang et al., 2007;Flynn, Wu & Melnyk, 2010;Narasimhan & Kim, 2002;Topolsek, 2011;Zhao et al., 2011;Alfalla-Luque & Medina-López, 2009;Carter, Sanders & Dong, 2008;Kaynak & Hartley, 2008;Li, Ragu-Nathan, Ragu-Nathan & Subba Rao, 2006). Nevertheless, there is a slight bias in research, both empirical and conceptual, that has leant towards external rather than internal integration (Zhao et al., 2011). This is why there have been calls so that any future research takes into account the relationships between the different components of the supply chain integration and the effect that each one has on the performance indicators of the company (Chang et al., 2007;Narasimhan & Kim, 2002;Zhao et al., 2011).
To help with the development of the proposed future research, in this paper our objectives are the clarification of the constructs, the creation of a measurement scale for the components of the supply chain integration, the validation of these scales and a preliminary study on the effects of a variety of control variables (size of the plant, country, sector and degree of vertical integration) in the values of these scales.

Definitions of integration
According to  supply chain integration can be defined as: "the degree to which a manufacturer strategically collaborates with its supply chain partners and collaboratively manages intra-and inter-organization processes. The goal is to achieve effective and efficient flows of products and services, information, money and decisions, to provide maximum value to the customer at low cost and high speed." This is why it is so important to instil confidence amongst all the agents, building long-term relationships, frequent communication, share both profit and risk, and look for effective ways of sharing information, make joint decisions and resolve conflicts . There are two main types of integration: external integration and internal integration (Zhao et al., 2011;Chang et al., 2007;.
Internal integration refers to the degree to which a company can organise its practices, procedures, information, decisions and conduct in a collaborative and synchronised way between its different areas, to be able to comply with client requirements and effectively interact with its suppliers (Zhao et al., 2011;Topolsek, 2011;. External integration refers to the degree to which a company understands the need of its clients and collaborates with clients and/or suppliers to develop inter-organisational strategies and shared practices and processes, so that it manages to satisfy its clients' needs . External integration consists of integration with clients and integration with suppliers Zhao et al., 2011;Escrig Tena & Bou-Llusar, 2005).

Control variables for supply chain integration
The use of operational management practices in general, and supply chain integration in particular, are normally affected by national culture, meaning that it is quite common to come across research where the country in which the plant is located explains to a certain extent the degree of use of supply chain integration (Oliver & Delbridge, 2002;Hofstede, 1998;Zhao et al., 2011;Pagell, Katz & Sheu, 2005). Another variable that often comes up is the sector (MacDuffie & Helper, 1997;Bruce, Daly & Towers, 2004;Bayraktar, Jothishankar, Tatoglu & Wu, 2007;Oliver & Delbridge, 2002;Roth et al., 2008;Martinez Jurado & Moyano Fuentes, 2011). There are also references to the fact that integration is associated with the size of the company (Underhill, 2001;Roth et al., 2008;Zhao et al., 2011). And finally, the degree of vertical integration can affect the type and degree of supply chain integration (Roth et al., 2008;Hayes & Wheelwright, 1984).

Method
The aim of this paper is to test the psychometric properties of a questionnaire to identify four constructs of supply chain integration in industrial companies.
We begin looking at a reflective model, where the items of the scales are estimators conditioned by a construct that can not be directly observed. The items therefore reflect this construct and are interchangeable, with the result that any group of these items will provide an estimation equivalent to the phenomenon of interest (Hair, Anderson, Tatham & Black, 1999;Brown, 2006;Byrne, 2006;Baxter, 2009).
The test bank of items used to build the survey originate from earlier works (Roth et al., 2008). Of these, 4 items have been selected for each construct, aiming to ensure that they are representative of the theoretical definition, used in recent papers, and that they are not redundant, to avoid the survey being excessively long. The score of the scales is the total of the sum of the items (Table 1).
Our empirical analysis is based on the HPM project database, the data for which was collected during the third round of this project (2005-2007 timeframe) by an international team of researchers. As a whole, the international sample is made up of 266 plants across ten countries: Austria, Canada, Finland, Germany, Italy, Japan, Korea, Spain, Sweden and the USA. In each country, the plants were randomly selected from three industries: automotive components, electronics and machinery. A stratified sampling design was used to obtain an approximately equal number of plants for each industry-country combination. The items were targeted at plant accounting managers, direct labour, human resource managers, inventory managers, process engineers, plant managers, quality managers, supervisors and plant superintendents. Items are responded to by at least two different managers/workers in the plant. After that, all the responses for each item in each plant were averaged to obtain plant items scores.

It13
We encourage employees to work together to achieve common goals, rather than encourage competition among individuals. (Stank et al., 2001;Wong & Boon-Itt, 2008;Germain & Iyer, 2006;Giménez & Ventura, 2003) It14 Departments in the plant communicate frequently with each other. (Kim, 2009;Sanders & Premus, 2005;Stank et al., 2001;Vickery et al., 2003;Wong & Boon-Itt, 2008;Germain & Iyer, 2006;Giménez & Ventura, 2003) It15 Management works together well on all important decisions (Narasimhan & Kim, 2002 We will then carry out internal consistency testing to purify the items (inter-item correlations, Cronbach's alpha, squared multiple correlation, corrected item-total correlation). The set of items that pass the internal consistency testing will be analysed using exploratory factor analysis with maximum likelihood and varimax rotation, to verify if each of the items has high loads on the predicted scales, and with a multi-trait/multi-item analysis to see the discriminant validity . And finally, a confirmatory factor analysis will be carried out using robust estimators, which will allow us to check the convergent and discriminant validity of the scales. This model incorporates the correlations of all the scales amongst themselves, given that certain theoretical evidence would appear to show that there is a certain overlapping between the constructs and that their correlations should therefore be taken into consideration .

Results
Our  (Table 2). Following this, a multi-trait/multi-item analysis was carried out (Table 5). To pass the test, the difference between the corrected item-total correlation and the item correlation with other scales should be greater than 0.123 -2* standard error (Doval Dieguez & Viladrich Segués, 2011)-. Item05, earmarked following earlier analysis as potentially having problems, has more correlation to an access other than that of the one theoretically assigned to it and its correlation is not sufficiently different in the other two axes. It is therefore an item that could create issues during discriminant validation and will therefore be eliminated from the model.
Currently, items it04 and It11 have passed the test.
The results of the exploratory factor analysis with factor extraction techniques using the maximum likelihood method and Varimax criterion under orthogonal rotation (Table 6), indicate that the sampling adaptation index (0.821) and Bartlett's test of sphericity   The final step in the process was the carrying out confirmatory factor analysis to complete checking the convergent and discriminant validation of each scale. We start with the joint measurement model, which is the best representation of the theoretical model where the scales are interlinked . In the first version, two scales had 4 items, and the others 3 items. All the factorial loads were greater than 0.6 with the exception of two items (It04 and It11), which have been eliminated from the definitive version. In the definitive estimations are significant and the standardised factorial loads are all greater than 0.6 ( Figure   1). The extracted variance of the scales is between .45 and .56 and the compound reliability Cronbach's alpha are in all cases greater than the cut-off value of .70 (Table 7). These analyses confirm the convergent validity of the proposed scales. At the same time, the scales also pass the test of variance extracted compared to squared correlations and the confidence interval for correlations (Table 7).  There are no significant differences in the sub-samples based on its size or the level of vertical integration. The general benchmark can therefore be applied to these business sub-groups.
There are only significant differences by industry for the degree of customer integration between machinery and the other three sectors. Although the differences are significant for the sub-samples of each country, the number of companies available in each sample is two small to be considered representative and therefore does not require the benchmark to be broken down.

Conclusions
This research paper provides an overview of the latest chain supply integration scales and expresses the need to formulate measurement instruments that allow one to identify the degree of use of each of the four constructs in companies (internal integration, external integration, integration with clients and integration with suppliers).
Starting out with a set of items, created especially for this research, 4 scales are proposed, and are subsequently validated using a broad sample. generalization of other industrial sectors (given that the sample consists of companies from only three sectors); or that the range of responses are concentrated in the upper part of the scale. This behaviour could stem from the characteristics of the sectors chosen for the sample, in which case it would be desirable to test out these scales in the future using a broader sampling and with plants from different sectors. In this way, the benchmark could be extended to be able to analyse differences by country or by sector (if these were available). Developing similar scales focusing on service companies that have their own set of characteristics when it comes to understanding and applying supply chain integration would be required.
The outcomes of this paper have obvious academic implications as it responds to requests expressed in recently published articles in this field, which asked for a clearer and more concise designation of the supply chain integration measurement scales. In this way, more reliable and accurate data could be taken to analyse the relations between these constructs with other variables of interest to the academic and professional fields, such as for example the outcomes or production efficiency.
From a professional perspective, this paper contributes to providing scales that are valid as a diagnostic tool for best practices, as well as providing a benchmark with which to compare the score for each individual plant against a collection of industrial companies from the machinery, electronics and transportation sectors.