ABSTRACT
In Chapter 2 we presented three major theoretical board compositional and structural groupings that guide our analyses. These groupings, including board member demographics, board member achievement and eliteness (now split into two categories), and board member networking capacity, help us to model our theoretical targets. Indeed, in Chapter 7 we outlined specific target variables that we will use in observing impacts of our predictors, in line with the theoretical models. In this way, using an information-based approach to model selection, we chose an initial set of candidate models using the theories explicated in Chapters 3 through 6. Our predictors, then, are Year (Y), Industry (I), City (C), and Religious (or Faith)-Based (R). We also add the predictor Board Size (B), based on the theory of bureaucracy also introduced in Chapter 3. As a way to fine-tune the analysis, as well as advance our current knowledge of predictors of board structure, we have added two-way interactions to all of our models. These interaction effects allow for slopes for the Board Size variable that differ depending on year, city, or industry, and main predictor effects that differ depending on the level of another predictor (for example, the CY interaction corresponds to a city effect that is different in 1931, 1961, and 1991). In all cases, we use information-theoretic measures to simplify models, thereby focusing attention on the most meaningful patterns.
