Concluding Remarks
Causality is a fundamental characteristic of a good
theory, but the difficulty in inferring causality has
forced researchers to either infer causality from pure
theory (Carte and Russell 2003) or from longitudinal
(Granger 1986), experimental (Cook and Campbell
1979), or panel (Allison 2005) data. This paper is an
attempt to revive the pursuit of causality in structural
models from observational data in the IS literature
in particular and the social sciences in general,
and encourage IS researchers to bring causality considerations
back into IS studies. The proposed BN-LV
method aims to provide a tool for IS researchers to
better understand how causal relationships can be
inferred in structural models from observational data.
We hope the proposed data analysis method serves
as a modest starting point for enhancing methods
for inferring causality and building causal theories in
the IS literature. Given the enhanced sophistication of
IS research in terms of theory and methods, causality
can become an important consideration in the IS
literature.