ABSTRACT

This chapter describes the original basis of the Reticular Action Model (RAM) approach

to structural equation modeling (SEM). The RAM rules were initially presented

(1978-1981) to simplify and unify structural equation models based on path analysis

graphics. The mathematical representation of RAM is presented as a second-order

moment structure model that includes two parameter matrices, one of which is a

patterned inverse. The graphic representation of RAM is presented through a series of

definitions and axioms that provide a complete and concise isomorphism between

graphics and algebra. Comparisons to other traditional models, such as those of multiple

linear regression, path analysis, and factor analysis, show how these models may be

easily and economically represented using RAM rules. Two other general modeling

foundations, LISREL and COSAN, are presented as special cases and, somewhat

paradoxically, as generalizations of RAM. These results are then used to develop some

important technical features of the RAM rules, including efficient algorithmic estimation

procedures and the further development of statistical indicators. Issues of conceptual

representation are provided from a general systems perspective. Finally, some of the

current limitations and benefits of the RAM rules are considered.