STRUCTURAL EQUATION MODELLING OF FACTORS THAT INFLUENCE THE SHOPPERS PURCHASE DECISION THROUGH E COMMERCE

Size: px
Start display at page:

Download "STRUCTURAL EQUATION MODELLING OF FACTORS THAT INFLUENCE THE SHOPPERS PURCHASE DECISION THROUGH E COMMERCE"

Transcription

1 27 STRUCTURAL EQUATION MODELLING OF FACTORS THAT INFLUENCE THE SHOPPERS PURCHASE DECISION THROUGH E COMMERCE Dr. D Uday Kumar, Assistant Professor, New horizon College of Engineering, Bangalore, India Dr. K Venugopal Rao, Professor, SKIM, Sri Krishnadevaraya University, Ananthapur, India. ABSTRACT E-commerce is buzzing word in today s generation where a rigorous study required last one decade there is an exponential growth in e-commerce industry so the aim of the present study is to develop a model of various factors that influence the shoppers purchase decision through e-commerce. Study was undertaken at two tier cities in Andhra Pradesh i.e. Rajahmundry, Vijayawada & Tirupati cities. Structural equation modeling was used to develop and test a model for the study. The data is collected primarily from a sample of 1200 shoppers. The findings show that there are 4 major constructs i.e. economic, product, convenience and credibility that has the positive influence on shoppers purchase decision through e-commerce and the model is validated by assessing different fit indices. Keywords: E-commerce, Structural equation model, constructs and fit indices.

2 28 Introduction: Internet and e-commerce had followed, inevitably, a similar road since these concepts cannot be mutually excluded one from the other one. Innovations in the field of Internet technologies have had instant repercussions in the online business world. From a simple usage having a regional origin located in the United States of America, the phenomenon of electronic commerce has seen a rapid spread globally, according to the innovations related to Internet technologies. The projection of electronic commerce is in perfect accordance with the development stage of the real economy. Definition of E-Commerce: According to IBM, which has defined this concept in 1997,''e-business can be the key in transforming business processes using Internet technologies.'' Analyzed from a general perspective, e-business may gather under its umbrella three major components: The human dimension that includes processes and activities related to research, development, marketing, manufacturing, logistics, management, etc. The technological-only component, akin to information-related technologies. Commercial or e-commerce component, taken as a whole and perceived primarily as a phenomenon having a different meaning that the term known e-commerce, basically the purchase of goods and services through Internet technologies. Review of Literature: Electronic commerce (e-commerce) is the buying and selling, marketing of products and services, and providing information via computer networks (O'Brien & maracas, 2006). Many companies are now engaged in or sponsor three basic categories of e-commerce applications, i.e., e-commerce business-to-consumer, business-to-business, and consumer-to-consumer (O'Brien, 2005). Lee et al. (2011) conducted a study in Malaysia in 2010; the instrument used was questionnaire survey using seven Likert scale. Data collection method used was a snowball effect for 102 respondents. Through regression and correlation analysis, it was found that perceived value, perceived ease of use, perceived usefulness, firm's reputation, privacy, trust, reliability and functionality have a significant linear relationship to online repurchase intentions. Delafrooz et al. (2011) who conducted a study in Malaysia with 370 respondents obtained from a private university in Selangor by using path analysis found that the trust and consumer attitudes have the strongest direct effect on buying online intention. While the utilitarian orientation, convenience, price, broader product selection, and earnings have also strong indirect effect on online shopping intentions through attitudes toward online shopping as mediation. Kwek et al. (2010) conducted a study of 242 respondents in Malaysia by using multiple regression analysis and found that impulse buying intentions, consumer orientation to quality, customer orientation to brands, online trust and online purchasing experience is positively related to prior purchase intention of online customers. Shergill & Chen (2005) collected data by conducting surveys via randomly to 102 respondents in New Zealand. Then, using exploratory factor analysis and ANOVA found that the site design, site reliability, customer service sites, and site safety are the four dominant factors that influence consumer perceptions of online purchases. Objectives of the Study: To identify the various factors and fit a model by using Structural equation modeling that influences the shoppers purchase decision through e-commerce. Research Methodology: Sample size: The researcher selected online shoppers in the cities of Rajahmundry, Vijayawada and Tirupati as a sample for the study.

3 29 Name of the city No. of frequent buyers through online Sample size Rajahmundry Vijayawada Source: compiled from websites Tirupathi Total: 1200 Frequently buying people are taken as respondents. Convenient sampling method has been adopted for data collection. Data collection: Both primary and secondary data has been collected for the purpose of the study. The survey method has been used to gather primary information for the study. The required data has been collected from the sample respondents with the structured questionnaire designed for the purpose. The secondary data has been collected through books, journals, magazines, internet, articles related for the study. Questionnaire: Based on the objective of the study, close ended questionnaire is prepared. Likert Scale method has been used, respondents have been asked to state their level of agreement or disagreement on 5 point scale where 5= strongly agree; 4= agree; 3= neutral; 2= disagree; 1= strongly disagree. Hypothesis of the Study: H 1 : Economic factor does not have positive influence on shopper buying decision through e-commerce H 2 : Product factor does not have positive influence on shopper buying decision through e-commerce. H 3 : Convenience factor does not have positive influence on shopper buying decision through e-commerce. H 4 : Credibility factor does not have positive influence on shopper buying decision through e-commerce Data Analysis: Initially researcher conducted exploratory factor analysis to extract the major factors and later conducted the structural equation modeling to develop and validate the model. Results and Discussion: The researcher has applied the exploratory factor analysis to study the various observed variables that influence the shoppers to opt for online shopping. There are 16 variables are considered for the study and 4 major factors was extracted from the study. The following figure-1 shows the 4 factors of the study. Figure 1 Economic Convenience Factors influence the purchase decision through e-commerce Credibility Product

4 30 Conducting Structural Equations Modelling (SEM) of factors influencing the purchase decision through e- Commerce: After applying the exploratory factor analysis, the model fit and hypothesis have been tested by using structural equations modelling (SEM) which is basically a combination of confirmatory factor analysis (CFA) and linear regression. Reliability Analysis of latent variables: Table 1: The cronbach alpha for the latent constructs Name of the construct Number of variables Cronbach s Alpha Economic Product Convinence Credibility Table 1 portrays the cronbach s alpha of four latent variables, that measures the consistency of the scales. Cronbach's alpha is a function of the number of items in a test, the average covariance between item-pairs, and the variance of the total score. Studies say that cronbach s alpha greater than or equal to 0.70 is said be good reliable and cosistent, the constructs of the study is having a cronbach s alpha economic with 0.992, product with 0.932, convinence with and credibility with 0.949, all the four constructs are having cronbach s alpha more than 0.9 which is greater than threshold and this shows the greater reliabilty and strong consistency among the variables. This is highly supportive for the conduction of structural equations modelling. Defining the Individual Constructs: Data have been analyzed using Structural Equations Modeling (SEM) with Maximum Likelihood Estimation (MLE). The first step under SEM is to define the individual constructs also called as factors or latent variables; four constructs from sixteen variables have been extracted by using exploratory factor analysis. Exploratory factor analysis has been conducted with Principal component analysis with a vari-max rotation method. Five variables are grouped under one construct named as Economic factor, four variables are grouped under one more construct named as product factor, four more variables are grouped under one construct named as convenience factor and there variables are grouped under final construct named as credibility factor. Specify the Measurement Model: The second step in structural equation modeling is to specify the measurement model where the indicators or observed variables are measured. This involves the assignment of the relevant measured variables to each latent construct. Measurement model ensures that predetermined indicators or observed variables are indeed reliable measures of the latent variable they represent. This measurement model were implemented using Confirmatory Factor Analysis (CFA), that is, to confirm the reliability of the measures and to determine the weight or contribution of each indicator to the latent variable it represents (factor loadings). The measurement model is usually represented by a path diagram figure 2 depicts the measurement model with four correlated constructs with each construct having five, four, four and three indicators or measured variables. Below Figure explains the complete measurement model; the assignment of measured variables to each latent construct is graphically equivalent to drawing arrows from each construct to the measured variables that represent that construct. The degree to which the each measured variable is related to its construct is represented by that variables loading, loadings with more than 0.9 explains the strong relationship between the observed variable and its respective construct. Figure 4.3 portrays the relationship between the observed variables with its construct with a single arrow, the values on the arrows represents the loadings where all the observed variables is having more than 0.9 as its loadings, this shows there is a strong explanatory relation between the observed variables with its constructs

5 31 Figure 2: Path diagram of Measurement Model The following is the description of the model for latent variable economic factor (EC): XX i = λ i EC + e i XX 1 = 0.97EC XX 2 = 0.98EC XX 7 = 0.89EC XX 10 = 1.02EC XX 12 = 1.00EC Whereas, XXi = Indicators of economic factor λ i = Factor loadings e i = Error terms The following is the description of the model for latent variable product factor (PR): XX i = λ i PR + e i XX 4 = 1.09PR XX 5 = 0.99PR XX 6 = 0.69PR XX 13 = 1.00PR Whereas, XXi = Indicators of product factor λ i = Factor loadings e i = Error terms The following is the description of the model for latent variable conveniencefactor (CV): XX i = λ i CV + e i

6 32 XX 3 = 0.97CV XX 11 = 0.98CV XX 14 = 0.92CV XX 16 = 1.00CV Whereas, XXi = Indicators of Convenience factor λ i = Factor loadings e i = Error terms The following is the description of the model for latent variable credibility factor (CR): XX i = λ i CR + e i XX 8 = 1.29CR XX 9 = 1.31CR XX 15 = 1.00CR Whereas, XXi = Indicators of Convenience factor λ i = Factor loadings e i = Error terms From the above measurement model the researcher has derived the following relationship between latent variables and exogenous variable. OPD (Exogenous variable) = β 1 EC + β 2 PR + β 3 CV + β 4 CR + e Where, OPD = Online purchase decision EC= Economic PR = Product CV= Convenience CR = Credibility e = Disturbance error β 1, β 2, β 3, and β 4 = Regression parameters OPD = 0.84E P C CR Assessing the measurement model fit: Goodness of fit means how well the measured model reproduces the covariance matrix among the indicator items that is how similar is the estimated covariance of the indicator variables to the observed covariance in the sample data. There are various fit measures to assess the model namely; absolute fit, incremental fit and parsimony fit indices. Assessment of model by absolute fit indices: Measures that are commonly used to assess the model are goodness-of-fit index (GFI) and adjusted goodness-of-fit index (AGFI). The GFI is a measure of absolute fit whereas the AGFI accounts for the degrees of freedom in the model. Values more than 0.90 is considered to be a good model fit and values more than 0.95 is absolutely good fit. Table 2: The GFI and AGFI Model RMR GFI AGFI PGFI Default model Saturated model Independence model Table 2 depicts the values of GFI and AGFI for the default model, it is observed that the GFI value is and

7 33 AGFI value is 0.908, which is greater than the threshold value and its fits the model, therefore by the absolute fit indices the goodness-of-fit index is supportive and can considered to be a good model fit. Table 3: Table showing the Chi-square statistics Model NPAR CMIN DF P CMIN/DF Default model Saturated model Independence model Table 3 shows the chi-square statistics, a chi-square test provides a statistical test of the difference in the covariance at specified degrees of freedom. Hence the lower the chi-square value is the better model fit, where in the above table CMIN is referred as chi-square value which is which is higher and based on chi-square statistics model is moderately fit. Table 4: Table showing the RMSEA Model RMSEA LO 90 HI 90 PCLOSE Default model Independence model Root mean square error of approximation (RMSEA) examines the difference between the actual and predicted covariance i.e. residual and specifically the square root of the mean of the squared residuals. Lower RMSEA values indicate better model fit, RMSEA value less than or equal to 0.08 is considered good model fit. Table 4 shows the RMSEA value of which is lesser than the threshold value; therefore it is observed that the measurement model is good fit. Assessment of model by Incremental fit Indices: Normed fit index (NFI) and comparative fit index (CFI) are also widely used model fit measures, and these represent incremental fit indices in that the specified model is compared to the null model in which the variables are assumed to be uncorrelated. Table 5: Table showing the NFI, TLI and CFI Model NFI Delta1 RFI rho1 IFI Delta2 TLI rho2 CFI Default model Saturated model Independence model Normed fit index (NFI) is a ratio of the difference in the chi-square value for the proposed model and the null model divided by the chi-square value for the null model i.e. as the chi-square value for the proposed model approaches zero, NFI tends to be a perfect fit of 1, the value of NFI greater than or equal 0.90 is said to be acceptable fit, hence in table 5 the NFI value for the default model is which is greater than threshold value, it is good model fit. Comparative fit index(cfi) is related to NFI and factors in degrees of freedom for model complexity, CFI varies from 0 to 1, and values of 0.90 or greater are associated with good model fit, Table 4.30 depicts the CFI value is which is greater than acceptable value, therefore model is said to be good fit. The Tucker Lewis index (TLI) is conceptually similar to CFI, but in not normed and so the values can fall outside of the 0 to 1 range. Models with good fit have a TLI value that is close to 1. Table 5 showing the TLI value which is near to 1 which is a good measurement model. Assessment of model by parsimony fit indices It should be emphasized that parsimony fit indices are not appropriate for evaluating the fit of a single model but are useful in comparing models for different complexities. The parsimony goodness-of-fit index (PGFI) adjusts the goodness-of-fit index by using parsimony ratio.

8 34 Table 6: Parsimony-Adjusted Measures Model PRATIO PNFI PCFI Default model Saturated model Independence model The parsimony normed fit index (PNFI) adjust the normed fit index by multiplication with parsimony ratio higher the PNFI is good model fit, hence table 6 portrays the parsimony normed fit index(pnfi) and parsimony comparative fit index (PCFI) is and respectively which is greater than threshold value hence it is good model fit. Therefore from the above analysis by using all three types of measures, the model is said to be fit and all the four constructs is having a positive influence on the shoppers purchase decision through e-commerce. Conclusion: From the above discussion it can be inferred that major four factors have huge impact on shoppers online purchase decision, the results of structural equation model fits a good model and proves the hypothesis that all the four factors is having the positive influence on the shoppers purchase decision through e-commerce. References: [1] O'Brien, J. A., & Marakas, G. M. (2006). Management Information Systems. 7th Edition. New York: McGraw Hill. [2] O'Brien, J. A. (2005). Introduction to Information Systems. 12th Edition. New York: McGraw Hill. [3] Lee, C. H., Eze, U. C., & Ndubisi, N. O. (2011). Analyzing key determinants of online repurchase intentions. Asia Pacific Journal of Marketing and, 23 (2), [4] Delafrooz, N., Paim, L. H., & Khatibi, A.(2011). Understanding consumer s internet Purchase intention in Malaysia. African Journal of Business Management, 5 (3), [5] Kwek, C. L., Tan, H. P., & Lau, T. C. (2010). The Effects of Shopping Orientations, Online Trust and Prior Online Purchase Experience toward Customers Online Purchase Intention. International Business Research, 3 (3) [6] Shergill, G. S., & Chen, Z. (2005). Web-based Shopping: Consumers Attitudes towards Online Shopping in New Zealand. Journal of Electronic Commerce Research, 6 (2), [7] Schiffman, L.G., Sherman, E., & Long, M.M.,"Toward a better understanding of the interplay of personal values and the Internet", Psychology & Marketing, 20 (2), pp , [8] Shergill, G., Chen, Z., (2005), Web Based Shopping: Consumers Attitude towards Online Shopping in NewZealand, Journal of Electronic Research, Vol. 6, No. 2, pp [Online] Available: [9] Solomon, M. R., Consumer behavior, New York, NY: Prentice Hall, [10] Sorce, P.; Perotti, V.; Widrick, S., Attitude and age differences in online buying, International Journal of Retail & Distribution Management, Vol. 33, No. 2/3, pp , [11] Sultan, F., Henrichs, R.B.,"Consumer preferences for Internet services over time: initial explorations", The Journal of Consumer Marketing, 17(5), pp , [12] ADIL BASHIR, Consumer Behaviour towards Online Shopping of Electronics in Pakistan, Thesis, winter 2013, MBA International Business Management. [13] Dr.payal upadhyay and Jasvinder Kaur, Analysis of Online Shopping Behaviour of Customer in Kota City, International Journal in Multidisciplinary and Academic Research (SSIJMAR), Vol. 2, No. 1, January- February (ISSN ). [14] Dr. Ravindra P. Saxena, Shoppers Attitude towards Online Retailing: An Exploratory Study Performed in the Context of United Arab Emirates (UAE), ravindrasaxena@uowdubai.ac.ae ****