Abstract
The social scientists work on economic problems associated with attitudes and perceptions of respondents to a survey on a given issue naturally face several challenges with regard to quantification of which to develop estimable variables to be used in further analyses. This article explores, using the special case of economic incentives for Sri Lankan agri-food processing firms to adopt enhanced solid waste management practices in the firm, the outcome of statistical methods employed to overcome such empirical issues, including: (a) “Mutual Exclusivity” and “Endogeneity” of incentives, i.e. prevalence of an individual incentive as an element of a system; (b) “Subjectivity”, i.e. the management of firm perceives unpredictably on these incentives in terms of potential benefits and costs to the firm, and (c) “Unobservability”, i.e. the management cannot directly observe the nature of incentives prevailing at the firm level. It uses the Structural Equation Modeling techniques with the aid of Analysis of Moment Structures (AMOS) statistical package to overcome these issues, where a family of statistical models that seek to explain the relationships among multiple variables were formulated by combining a Measurement Model [commonly referred to as Confirmatory Factor Analysis (CFA)] with Structural Model into a simultaneous statistical test. The outcome of analysis, which used data collected from 325 firms by means of a questionnaire-based survey comprising of 9 Constructs/latent variables (i.e. incentives considered in the analysis) and 51 Indicators (attitudinal statements), facilitate deriving an Incentive Index for each incentive reflecting its relative strength at the level of firm, and in turn, to use as explanatory variables in modeling. Sri Lankan Journal of Applied Statistics, Volume 13 (2012), p. 15-37 DOI: http://dx.doi.org/10.4038/sljastats.v13i0.5122
Highlights
The social scientists work on economic problems associated with attitudes and perceptions of respondents to a survey on a given issue naturally face severalSri Lankan Journal of Applied Statistics, Volume 13 (2012), p. 15-37challenges with regard to quantification of which to develop estimable variables to be used in further analyses
In order to overcome these difficulties and to develop such variables, we have decided to use the Confirmatory Factor Analysis (CFA) techniques [i.e., a multivariate data analysis technique that comes under Structural Equation Modeling (SEM)] for the 9 individual incentives (j = 1, 2...9), which we have summarized below
Following the standard guidelines for constructing a validation item for the CFA, we included 9 validation items in the questionnaire to represent corresponding individual incentives.Given two or more Constructs and two or more ways of measuring each, we can expect a high correlation between these two different measures when they are used to evaluate the same Construct, but a low correlation between these measures when used for different Constructs, or in statistical terms this satisfies the condition of Convergent Validity and Discriminant Validity (Campbell and Fiske, 1959)
Summary
Challenges with regard to quantification of which to develop estimable variables to be used in further analyses. SEM provides a useful way in which to determine whether observed data concur with a priori hypotheses on the structure of incentives (Hughes et al, 1986; Joreskog and Sorbom, 2001). It has the advantage of providing a method for dealing with multiple and inter-related dependence relationships, while providing statistical efficiency and to assess directly unobservable concepts for which respondents possess subjective assessments in terms of a number of observable components (Hair et al, 1998). SEM has been used in previous empirical studies of consumer and managerial behavior (see, for example Henson and Traill, 2000; Nakamura et al, 2001) with great success
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have