Home / Market Development

How to Match Data for Two Data Sets of Manufacturers – Business Customers in the Consumption Channel

Social science research has turned to second-generation statistical techniques called structural equation modeling (SEM). The popularity of structural equation modeling (SEM) has grown due to the need to test theories and refine research concepts (Rigdon, 1998).

There are two types of SEM models including covariance-based approach (CB SEM) and variance-based partial Least Square technique (PLS SEM). CB SEM is used to test theories by determining which model best estimates the covariance matrix for sample data, commonly used in AMOS and LISREL software. Meanwhile, PLS SEM operates like multiple regression analysis (Hair et al., 2011), used by SmartPLS software. This means that PLS SEM is suitable and valuable for research with exploratory purposes (F. Hair et al., 2014). This reason is consistent with many previous studies.

Recently, PLS-SEM has gained much attention in various research fields such as marketing, strategic management (Hair et al., 2012), management information systems (Ringle et al., 2012), operations management (D.X. Peng & Lai, 2012), and accounting (L. Lee et al., 2011). Much of the increased use of PLS-SEM is due to its ability to handle modeling problems that are common in social sciences, such as nonnormal data characteristics and high model complexity (Hair et al., 2014).

PLS-SEM is used as a preferred option for handling small samples (Hair, Hult, et al., 2017; Hair et al., 2014). The sample size used in this study is 78. Because the entire Mekong Delta construction materials manufacturers are small, the number of sample sizes collected is the entire population with 78 owners or senior managers of small and medium-sized enterprises. Furthermore, other reasons that Hair, Hollingsworth, et al. (2017) stated such as non-normally distributed data, maximizing the variance explained for endogenous variables, multiple interaction terms with each other (complex model), and a large number of observed variables, scales derived from previous research, explaining an outcome of interest, identifying relationships, having both second-order and first-order research variables in the research model. These reasons are also very suitable for the thesis to use PLS SEM in testing the research model.

Thus, PLS SEM 3.3.3 is used in this thesis to examine the supporting factors of market orientation impact, innovation capacity of manufacturers - business customers in the consumption channel to develop the construction materials market and for the following reasons: (1) to avoid problems related to small sample size; (2) non-normally distributed data; (3) to estimate a complex research model, with mediating variables (MOC, IC, GS) and other latent variables interacting with each other; (4) to have both second-order research variables (MO, IC) and first-order variables (GS, MD) in the research model.

Maybe you are interested!

b) Steps to perform formal quantitative data analysis

Formal quantitative data analysis in the thesis is carried out through the following four steps:

How to Match Data for Two Data Sets of Manufacturers – Business Customers in the Consumption Channel

Step 1: Descriptive statistics

Descriptive statistics can be defined as a method related to data collection, summarization, presentation, calculation, different characteristics to reflect the general interview subjects. Statistical tools used in data analysis include statistical tables, frequency distribution tables, comparison of absolute numbers and relative numbers. Statistical tables present collected data and information (research results) as a basis for analysis and conclusions. Frequency distribution tables are tables summarizing data arranged into different groups based on the frequency of occurrence of objects in the database to compare proportions and reflect data.

The above descriptive statistics are used to describe the characteristics of the survey subjects of 236 enterprises, with 78 enterprises being manufacturers and 158 enterprises being corporate customers with criteria such as education level, expertise of respondents, type of enterprise, field of operation of the enterprise, enterprise size, contents related to production and consumption of the enterprise.

Step 2: Evaluate the measurement model

Evaluation of the measurement model is based on three values ​​including composite reliability, convergent validity and discriminant validity.

Composite reliability measures the reliability of a set of observed variables measuring a concept (factor) and the reliability coefficient (Cronbach's Alpha - CA) measures the internal consistency across the set of observed variables of the responses. The reliability of the observed variables must have a reliability coefficient greater than or equal to 0.6 to meet the reliability requirements (Hulland, 1999), the Composite Reliability coefficient - CR must be greater than or equal to 0.7 to meet the composite reliability (Fornell and Larcker, 1981).

- Convergent validity is used to assess the stability of the scale. The scale achieves convergent validity when the standardized weights of the outer loading coefficients of each observed variable on the factor are greater than or equal to 0.7 and statistically significant (p < 0.05) and the average variance extracted (AVE) value reflects the amount of common variance of the observed variables explained by the latent variable must be greater than or equal to 0.5 to confirm convergent validity (Henseler et al., 2009).

- Discriminant validity measures the discriminant validity to ensure the difference, no correlation between the factors used to measure the factors. To measure discriminant validity, the Fornell – Lacker criterion should be used: the square root of AVE of each measured factor is greater than the latent variable correlations between that factor and other factors, showing the discrimination and reliability of the factors (Henseler et al., 2009).

- Collinearity assessment: To assess the collinearity of variables, PLS-SEM uses the Variance Inflation Factor (VIF). When VIF exceeds 10, it is a sign of multicollinearity (Hair et al., 2014), and the variable must be removed from the model. According to Hair et al. (2011), a VIF value < 5 means that there is no multicollinearity between the research variables and the model will continue to be analyzed.

Step 3: Evaluate the structural model

After the measurement model meets the requirements, evaluate the structural model through the following criteria:

- The overall coefficient of determination R 2 (R – square value) is an index to measure the level of fit to the data model or the explanatory ability of the model. Hair et al. (2014) stated that the R 2 (R – square value) in the PLS SEM measurement model has values ​​of 0.67; 0.33 and 0.19, which mean strong, medium and weak, respectively.

- Assessing the structural model path coefficient (Path Coefficient): shows the level of impact of variables on each other. The path coefficients have a standardized value approximately between -1 and +1. In which, estimated path coefficients with positive values ​​indicate a positive relationship and vice versa for negative values ​​(-) indicate an inverse relationship; the closer to 1, the stronger the relationship. Low values ​​close to 0 are often not statistically significant (Hair, Hult, et al., 2017). In addition, it is necessary to consider the p_value of all path coefficients in the model when performing the bootstrapping procedure, accordingly, p_value < 0.05 (or p_value < 0.1) will conclude that the considered relationship is statistically significant at the 5% (or 10%) level (Hair, Hult, et al., 2017).

Step 4: Bootstrapping Test – Test the reliability of the SEM model.

After completing the estimation of the research model, re-evaluate the reliability of the research model estimate to extrapolate to the population. Schumaker and Lomax (2004) proposed the bootstrapping testing method as a repeated sampling method, in which the initial sample acts as a crowd. This method uses an approach that is not based on the interaction between variables and factors to predict the accuracy of relationships in PLS. With the bootstrapping technique, the recovered sample can be considered as a population, N sub-samples in the population are created by sampling with changes in observed values ​​in the initial sample size (in the study N = 50). Then, the relationships begin to be predicted for each newly created sample. The distribution of predictions from the M samples is created to calculate the t_value or p_value of the relationship. After bootstrapping, it is necessary to re-examine the significance of the relationships, the total direct and indirect effects at the 5% or 10% significance level (suitable for exploratory studies). In addition, the suitability of the model to the research area is measured by the standardized root mean square residual (SRMR). According to Hu and Bentler (1999), the SRMR index must be less than 0.08 or 0.1. In addition, Henseler et al. (2014), Lowry and Gaskin (2014) also stated that the SRMR index is a goodness of fit index of the PLS-SEM model that can be used to avoid parameter bias in the model.

3.4 How to match data for two sets of manufacturer – business customer data in the consumption channel

To perform data analysis, after the data is collected, each data set is checked and cleaned, then data matching will be carried out.

Many researchers have performed data matching from two data sets collected from two different survey subjects (!!! INVALID CITATION !!! (Francescucci et al., 2018; Kibbeling et al., 2013; Langerak, 2001; Siguaw et al., 1998)). In this thesis, the author inherits the data matching method of previous studies. Before testing the relationship between market orientation and innovation capabilities of manufacturers and business customers with product market development of manufacturers

In order to produce under government support regulation, it is necessary to match two sets of data of manufacturers: VLXKN business customers at a ratio of 1:2 into a synthetic data set. The implementation is as follows:

Step 1: Calculate the average market orientation score of business customers. For example, the market orientation assessment of the 1st and 2nd business customers of each manufacturer will be averaged.

Step 2: Proceed to match the market orientation data, the innovation capacity of the manufacturer and the market orientation of the business customers into a composite file As mentioned in section 3.3.2.2 Selecting a sample for the main quantitative research

In this way , two data sets from the manufacturer and the enterprise customer are merged into

a data set to conduct the tests. This data set is completely acceptable and reliable because the manufacturer and the business customer are each other's major customers. This inherits the way of setting up the survey according to the partnership relationship and consumption channel of previous studies (Francescucci et al., 2018; Deshpandé et al., 1993; Kibbeling et al., 2013; Kotabe et al., 2003; Langerak, 2001; Siguaw et al., 1998). In addition, Kibbeling et al. (2013) also emphasized that further studies should consider more cases instead of just one manufacturer and one of their major customers to increase the reliability of potential consumption channel effects. In this thesis, the representative of the manufacturer of construction materials answered the survey introducing information of at least 3 of his own business customers (construction contractors) in 2018-2019 to serve the next survey.

Conduct data matching, each manufacturer has its own market orientation and has an average assessment of the market orientation of its business customers ( See Appendix 13B for the list of manufacturers and their respective business customers ). The simulation of data matching is as follows:

Manufacturer 1: Market Orientation

Average market orientation of manufacturer's business customers 1

Manufacturer 2: Market Orientation

Average market orientation of manufacturer's business customers 2

Manufacturer 78: Market Orientation

Average market orientation of manufacturer's business customers 78

Figure 3.2: Simulation of thesis data matching

Source: Author's proposal


After completing the data set into a file, perform data analysis as mentioned above.

Chapter 3 Summary

From the research objectives, the author designs the research process and selects research methods. Three stages are carried out in the research, including building a theoretical research model from theoretical research and application context, building a complete scale and conducting quantitative research. Two research methods are applied in the thesis, including qualitative and quantitative. The author uses the PLS-SEM model through the SmartPLS tool to test the proposed model.

CHAPTER 4

RESULTS AND DISCUSSION

Chapter 4 presents and discusses the research results based on the analysis of collected data. In this chapter, the author presents the following contents: (1) Introduction to the development of the non-fired construction materials industry, (2) Results of analysis of characteristics of manufacturers and customers of non-fired construction materials enterprises, (3) Results of evaluating the measurement research model and structural model, (4) Conclusion and discussion of the research hypothesis through PLS SEM technique (SmartPLS 3.3.3 software) from evaluating the measurement model, evaluating the structural model, (5) Discussion of the test results in the context of non-fired construction materials.

4.1 Overview of the unburnt construction materials market

4.1.1 Characteristics of unburnt construction materials

The qualitative research results have determined that VLXKN in this thesis are considered as materials made from a mixture of binders, aggregates, additives and water.

etc. according to different technologies without going through the firing stage, used in building house walls, building fences, building park perimeters, sidewalks, paving yards, roads, bridge decks, sewers, paving floors, warehouse floors, yards, and many other applications in construction and life today. Construction materials are input materials through surveying, designing, constructing and supervising activities, which will have output products such as residential houses and non-residential houses in the industrial and civil construction industry. The output products of the construction industry are classified according to the purpose of use, including: residential houses, non-residential houses and infrastructure. In addition, residential houses and non-residential houses can also be divided into civil (including houses, hotels, offices, commercial centers...) and industrial (industrial parks, factories, plants...).

Construction materials are the main raw materials that make up construction works. During the circulation process on the market, construction materials are mostly goods with large volume, bulky when transported, stored, and purchased; some types are easy to cause dust, easy to burn, affecting the environment and social management order, especially in urban areas (Loc, 1995). Construction materials are basic construction products such as cement, bricks, concrete and aggregates, namely sand, stone and gravel (Sinh, 2019). Therefore, VLXKN has the common characteristics of construction materials and belongs to the group of supporting industries. Therefore, the characteristics of this product market include: (1) The supply of construction materials for the purpose of producing final products such as housing, non-residential housing and infrastructure; (2) This supply is mainly met by the system of SMEs with technological level, creating products with high precision, implementing contractual commitments with customers in a standard manner; (3) the main customers of supporting industries are assemblers, not as wide as the production of products for final consumers. The market for construction materials is narrower. This is the biggest difficulty in developing this industry. However, the production of construction materials becomes attractive and relatively stable if the construction materials manufacturing enterprise finds long-term customers (Hung, 2015).

4.1.2 Background of the non-fired construction materials industry

According to the socio-economic development strategy report 2011 - 2020, the construction industry is an economic sector with an important strategic position and role in the process of economic construction and development. The total production value of the Vietnamese construction industry achieved a growth rate of 9.2% in 2019, the urbanization rate of the whole country reached 39.2% (up 0.8% compared to 2018), the total production value of the whole construction industry reached 358,684 billion VND, contributing 6.76%, higher than the growth rate of 2011, 2012 and 2013 in the period 2011-2020 in the GDP structure of the whole country (Government, 2020; General Statistics Office, 2020). The total number of construction enterprises has grown significantly, both in terms of input (raw materials, labor and machinery and equipment) and output (housing, non-residential housing, infrastructure) from 42,868 enterprises in 2010 to nearly 80,000 enterprises in 2018, with an average enterprise growth rate of about 7.3%/year (Vietnam Construction Association, 2018).

Vietnam's construction industry is developing strongly in many areas such as construction technology, construction project management, construction materials, architecture and construction planning, urban and housing development. Construction materials account for 70% of the input value contributed to the construction industry value chain (Ministry of Construction, 2020d). Most of the construction materials that need to be produced are cement, steel, bricks, etc. The construction materials industry is a key industry, developing with an average annual growth rate of about 8% to 10% in line with the orientation of the Vietnam industrial development strategy to 2025, with a vision to 2035 of the Vietnamese Government (Ministry of Construction, 2020, p.7).

Innovation in the production and consumption of construction materials, specifically unburnt construction materials (UCMs), is one of the long-term solutions for economic development and environmental protection. Production and use of unburnt construction materials is a solution to reduce the risk of environmental pollution by reducing emissions (CO 2 , NO 2 , NO x ) and food impacts due to loss of agricultural land when producing fired clay bricks, utilizing some types of industrial waste as raw materials (Ministry of Construction, 2020, p.7). Continuous innovation is an important factor for sustainable development of the construction industry (Czarnecki & Gemert, 2017). Therefore, new policies from the Government, Ministries and local authorities to encourage production establishments or enterprises to convert investment in the production of construction materials have just been implemented in the 10-year period (2010-2020) and continue to implement the strategy for developing Vietnam's construction materials for the period 2021-2030, with a vision to 2050.

4.1.3 Industry capacity scale

Vietnam's construction materials industry has made significant progress. In the period 2010 - 2018, investment in construction materials factories increased sharply. The design capacity accounted for about 8% in 2010, increasing to 30% of the total design capacity of construction materials. In terms of the construction materials industry, the Government's construction materials development program until 2020, the actual production scale of construction materials replacing fired clay bricks is set to reach 75% of the program's target - meaning that construction materials must account for 40% of construction materials by 2020. However, the data in Figure 4.1 shows that unburnt construction materials currently account for only more than 30% of the total construction materials. Thus, the implementation results are not yet

Comment


Agree Privacy Policy *