preview

Confirmatory Factor Analysis Paper

Better Essays

In order to examine the first research question, which focuses on whether the two sets of environmental behaviors are essentially the same or not they are systematically analyzed using confirmatory factor analysis. The foundation of this analysis relies on the fact that environmental behavior may take many forms and that those who are concerned about the environment may engage in some behaviors more so than other forms. This possible difference in engagement may be due to opportunity structures that help or hinder the feasibility of engaging. For example, if one does not have access to a curbside-recycling program, it would follow that the likelihood of engaging in this specific behavior would be lower than someone who did. This is just one …show more content…

This analysis focuses on the covariances for a number of reasons. First, few models had cross-loading indicators. Second, in order to test whether each set of indicators were related or not I thought it best to keep the direct influence of latent constructs from one set of indicators to the other latent construct at a minimum. Third, many of the suggested additional paths were within models, while only some were across models. Fourth, theoretically speaking, it made more sense to have the paths added between indicators rather than paths from one indicator for green consumerism to an indicator to traditional environmental activism. Moreover, this allowed for a more nuanced approach to understanding each set of indicator relationships. Lastly, the paths suggested between indicators change the model and in all cases resulted in models with few to no cross-loading indicators for the covariances accounted for the association between items and …show more content…

These included Akaike Information Criterion, the Bayseian (Schwarz) Information Criterion, the Tucker-Lewis Index, the Comparative Fit Index, the Standardized Root Mean Squared Residual, Root Mean Squared Error of Approximation, and the Co-efficient of Determination, which are all obtained via the ESTAT command. Each of these measures has their own criterion that demonstrates good model fit. I used these in order to determine which model fit the best when the modification indices above provided several possible final steps in creating the best overall fitting model. What I mean by this is that in some cases, the modification indices had two suggested paths to add and regardless of which one was added the model would fit the same, or nearly the same-hundredths of a point difference in Chi-Square value. Thus, I added each path in separate models, and utilized the above fit statistics to assess which path offered the best fitting model. In most, if not all, cases these measures were near identical. Thus, the final model and path selection relied on which path had the greatest substantive meaning to the analysis, while not compromising overall model fit. The analysis went through this process for each year and each latent construct

Get Access