Theoretical Statistics Analysis - Assignment Example

 Statistics Tools Adoption vs. Acceptance of E-commerce: Two Different Decisions Journal To test for reliability and internal consistency of the scales used, Cronbach alpha coefficient and composite reliability coefficient were used for the test. According to Nunnally (1978), a Cronbach alpha coefficient between 0.7 and 0.95 is reliable and that scales with Cronbach alpha less than 0.7 are not reliable. Cronbach alpha normally identifies scales/factors which can be omitted from the analysis so as to maintain reliability of the results. Another statistical too used were the multi-sample analysis and Lagrange multiplier test to test the difference between the attitudes of potential e-customers and experience e-customers and to confirm the formulated hypothesis. Chi-square tests are also performed to determine relationship between adoption and actual use. Chi-square tests are used in situations where the data can be put in contingency tables or in cases where there is need for test of goodness of fit. In this case, the chi-square statistics are computed and compared with the various probabilities for significance. According to Blanca, Julio & Jose (2009), to compare any differences that may exist between potential future users of e-commerce and by experienced e-commerce customers, random sampling of the participants is used. However, this goes against requirements of sampling that a sample needed to be un-biased. Although random sampling will ensure an un-biased sample, the whole population is not homogenous and is made up of people of different traits which should occasion stratified sampling. Stratified sampling should have been used as the whole population will be sub-divided into non-overlapping strata and then select a random sample from each stratum and combine to have an equally representative sample. Of the 805 participants in this survey, 580 were potential e-customers while the remaining 225 were experienced e-customers. There is the need to use binomial tests to test for statistical significance in the sample difference. Normally, a binomial test will test whether potential e-customers and experienced e-customers are distributed at a proportion of 0.5. Further, t-tests can be used to test the differences between the attitudes of potential e-customers and experience e-customers and to confirm the formulated hypothesis. But first there is need for normality tests to confirm whether the data is normally distributed so as to decide whether to use parametric or non-parametric independent samples t-test. T-tests would produce differences in samples. Another statistical tool that would have been used is descriptive statistics to show the sample profiles for the different aspects of the study. Descriptive statistics provides a bird’s eye view of the data and are easier to understand even by non-statisticians. Further, logistic regression would have done better to explain the odd ratios of adopting e-commerce by potential customers and also to show the trends in respect to years, income and gender etc. Another important test would have been multi-collinearity test to test for collinearity between the scales as collinearity affects negatively the model coefficients. Strategic value and electronic commerce adoption among small and medium-sized enterprises in a transitional economy Journal The aim of this paper was to determine the relationship between the perceptions of the strategic value of e-commerce and e-commerce adoption among 107 owners/managers of small and medium-sized enterprises (SMEs) in Ghana (Saffu & Walker 2008). The first statistical tool used was the test for outliers to determine extreme points which may sway the results negatively. Although the questionnaire was test-re-tested for reliability through a pilot survey, reliability analysis using Cronbach alpha coefficient was performed and the scales found to be reliable and valid. Descriptive statistics were also used to represent the sample profiles; means, frequencies and standard deviation. This test is good and shows the sample distribution and location of data as well as the amount of spread. Factor analysis was also carried out using the component matrix determining the extraction level. Factor analysis is used to reduce the variable or to detect the structure of data. Further, canonical analysis aimed at studying the existence of any interrelationships among the different sets of multiple variables was conducted (Saffu & Walker 2008). Also, correlations coefficient between the different variables were carried out and based on them, F-statistics were calculated and tested for significance. The following statistical tools could be applied to give more insights into the data. They include; parametric or non-parametric t-tests depending on the distribution of the data to determine whether there is a statistical difference between commerce adoption among small and medium-sized enterprises. Further, there is need for graphs/charts to represent the data in a more comprehensible manner for easier understanding. Discriminant analysis could have been used to determine which variable discriminates the other groups. Discriminant analysis is mostly used as it shows the variables which result into better prediction of the model. Lagrange multiplier test could also be used to test whether addition of a certain variable into a model without it would improve the model of not. This combined with adjusted R-squared could lead to the determination of the best fitting model (since the larger the R-adjusted, the better the model). There is need for binomial tests to test the representation of the sample for both gender and small and medium-sized enterprises. This will result into un-biased sample which will mean that the results can be generalized with ease as the sample will have attained the required scientific rigor. Further, logistic regression would be better in explain the effect on the odd of adopting e-commerce and also to show the trends in respect to number of employees and income etc. Another important test to be performed would have been multi-collinearity test to test for collinearity between the scales as collinearity affects negatively the model coefficients by deviating the error terms from homoskedasticity to heteroskedasticity thus violating the assumptions of econometrics (Creswell 2006). Thus, the estimates are un-biased and cannot be used to make inferences about the population. The relationship between buyer and a B2B e-marketplace: Cooperation determinants in an electronic market context Journal To understand the relationship between buyer and a B2B e-market place, Lancastre & Filipe (2006) employed stratified sampling where the entire population of respondents was stratified into strata and then simple random sampling carried out. The authors used sampling design tools to determine the size of the un-biased sample which could be employed in the whole survey. The formula; n = N / (1+N (e2)), (where N = total population of buyer and B2B e-market place while e = error term). This is a good practice in sample survey design and ensures generalizability of the end results. Sample profiling was also determined using descriptive statistics; frequencies and variations. This is encouraged in statistics research work as it shows data distribution. However, the problem with the representation of the results should be in tabular form for easier understanding and clarity (Curran & Perecman 2006). Further, tests for sample biasness were carried out to ensure un-biased representation which as earlier stated would lead to generalizability of the end results. Confirmatory factor analysis were further used to test for validity and reliability of the variables being measured and using Chi-square statistics tested for significance. Further used were the comparative fit index, incremental fit index, and the non-normative fit index and were meant for testing for discrepancies between the observed and predicted covariance matrices. Another statistical tool used is the prediction of estimates to establish commitments and testing for their significance. The following statistical tools could be applied to give more insights into the data. They include; parametric or non-parametric t-tests depending on the distribution of the data to determine whether there is a statistical difference between buyers among B2B e-marketplace. Further, there is need for graphs/charts to represent the data in a more comprehensible manner for easier understanding. Discriminant and factor analysis could have been used to determine which variable discriminates the other groups. According to Marczyk (2005), discriminant analysis is mostly used as it shows the variables which result into better prediction of the model. Lagrange multiplier test could also be used to test whether addition of a certain variable into a model without it would improve the model of not. This combined with adjusted R-squared could lead to the determination of the best fitting model (since the larger the R-adjusted, the better the model). Logistic regression would have done better to explain the odd ratios of relationship between buyer and a B2B e-marketplace s and also to show the trends in relationship. Another important test would have been multi-collinearity test to test for collinearity between the scales as collinearity affects negatively the model coefficients. Reference Blanca, H., Julio, J. & Jose´, M. (2009). Adoption vs. Acceptance of E-commerce: Two Different Decisions. European Journal of Marketing Vol. 43 No. 9/10, pp. 1232-1245 Creswell, J.W (2006). Qualitative Inquiry and Research Design: Choosing among Five Approaches, 2nd edition. Sage Publications, Inc Curran, R.R and Perecman, E. (2006). A Handbook for Social Science Field Research: Essays & Bibliographic Sources on Research Design and Methods. Sage Publications, Inc Lancastre, A. & Filipe, F.L. (2006). The relationship between buyer and a B2B e-marketplace: Cooperation determinants in an electronic market context. Industrial Marketing Management, 35, pp.774–789 Marczyk, G.R (2005).Essentials of Research Design and Methodology (Essentials of Behavioural Science). Sage Nunnally, J.C. (1978). Psychometric Theory, 2nd ed., McGraw-Hill, New York, NY Saffu, K & Walker, J.H. (2008). Strategic value and electronic commerce Adoption among small and medium-sized enterprises in a transitional economy. Journal of Business & Industrial Marketing, 23/6 (2008) 395–404 Read More