Influence Networks 10

variable weighting is captured within a regression model. In this model, behavioral intention is

regressed onto attitude and subjective norm, where *i *represents individuals and the slopes, ÃŸ

1

and

ÃŸ

2

, represent the expected change in behavioral intent associated with a unit increase in attitude

or subjective norm. These slopes represent the relative contribution, or weights, for attitude (ÃŸ

1

)

and subjective norm (ÃŸ

2

).

Behavioral Intent* *

*i*

= ÃŸ

0

+ ÃŸ

1

Attitude* *

*i*

+ ÃŸ

2

Subjective Norm* *

*i*

+ *r*

*i*

The original model is based on groups of individuals, but for the purpose of this study it

was necessary to derive weights for each individual within their respective academic department.

Recent developments in modeling procedures have made it possible to measure individual

variation between these behavioral antecedents (Hedeker, Flay, & Petraitis, 1996). Hedeker and

his colleagues derived individual level relative weights to empirically verify that the contribution

of each component varies across individuals (Hedeker et al., 1996).

Individual-level weights for attitude and subjective norm can be estimated, in principle, if

multiple observations are used as the random effect variable, either over time or across

behavioral observations. This study employed observations for a set of teaching behaviors as the

random effect variable. The first level of the multilevel model represents individual level

cognitions and behavioral intentions for a set of teaching behaviors. In the following formula,

teaching behaviors constitute the levels of the random-effect variable, which is represented by *j*

(Hedeker et al., p. 110):

Behavioral Intent* *

*ij*

= ÃŸ

0* i*

+ ÃŸ

1* i*

Attitude* *

*ij*

+ ÃŸ

2* i*

Subjective Norm* *

*ij*

+ r

*ij *

Level 2: Adding a Structural Predictor

The second level of this model contains structural centrality. Centrality appears at this

level because it is based on contextual information rather than individual level perceptions. Two