220050 SE SE Advanced Data Analysis 2 (2016S)

It is widely acknowledged among communication scientists, psychologists, and marketing researchers that empirical research including latent variables or constructs (e.g., customer satisfaction, consumer brand engagement, brand identification) enhance our understanding of complex social phenomena. This course seeks to familiarize participants with the nature, measurement and inclusion of latent measurement in research models that advance our knowledge in communication science. The topics discussed in the course are twofold:

In the first part - after reviewing the nature and role of latent variables - a broad introduction to measurement theory as well as alternative approaches for measuring constructs is provided. This section demonstrates how latent variables are created by researchers and addresses the procedures for developing multi-item scales that validly and reliably capture a construct's true nature. The course offers a user-friendly but profound insights to all stages of the scale development process and highlights key decisions as well as potential problems at each stage. It enables participants to experience measure development in action by discussing various issues which are illustrated with concrete examples of reflective scale development and formative index construction drawn from literature. Once the participants are familiar with the basic measurement principles, more advanced topics such as higher-order constructs and single-item measurement models are addressed.

The second part deals with the introduction of latent variables to structural models and the testing of hypothesized relationships among latent variables with empirical data. By doing so, the course gives an introduction to structural equations modelling (SEM) using the AMOS program. The discussion will include, amongst others, issues associated with the theoretical specification and graphical representation of a SEM model. Following the theoretical background, the course enables its participants to apply the AMOS program to estimate a model and to evaluate the model's fit along different criteria. Alternative strategies for model modification and cross-validation are also outlined along with detailed guidelines for setting up and interpreting the relevant input/output files of the AMOS program.

The course will take the form of interactive workshop sessions, placing particular emphasis on student participation. Theoretical discussion of key issues will be accompanied with examples taken from literature and practical exercises in the computer lab.

The course is designed for master students and assumes previous knowledge of empirical data analysis and statistics (including regression and correlation analysis). While the course includes some practical tutorials on the AMOS program in a computer lab, it is highly recommended that participants to install AMOS on their computer. Go to https://zid.univie.ac.at/software-shop/ for reduced software prices.