ABSTRACT
Chapter 2 discusses why screening your data is important before moving to the analysis stage. This chapter discusses in detail how to assess if you have respondent abandonment, if you have impermissible values, or if a respondent is “yea-saying”. Additionally, a discussion is presented on how to determine if you have missing data and how to address/impute missing data before your analysis.
The second half of the chapter introduces the concept of reliability and how to calculate reliability of indicators through the use of Cronbach’s alpha. The topic of identification in SEM is then addressed and how it relates to degrees of freedom. Further, an outline is given on how to calculate degrees of freedom for measurement and structural models. The chapter then explores the issue of sample size with SEM models and how big a sample is ultimately needed. Lastly, the topic of validity is introduced with a discussion of content validity, convergent validity, discriminant validity, and predictive validity.
