A Study of Senior Information Systems Managers' Decision Models in Adopting New Computing Architectures

Abstract

There has been considerable interest recently in the promise of new computing architectures such as the diskless computing architecture, which runs applications off a network. In previous theoretical work on Information Systems (IS) adoption, the question of whether classical diffusion variables determine the organizational adoption of IS with low knowledge barriers and low user interdependencies is still unresolved. In the practitioner literature, the discussion on new architectures has focused mainly on the costs of ownership of the architectures. This work proposes a novel methodology for IS adoption studies, using conjoint analysis. Issues such as data collection, data analysis, selecting scales and levels of predictor variables, construct validity, formulating testable hypotheses, and selecting an appropriate sample size are all discussed. As an example study, factors important to senior IS managers when deciding to adopt a computing architecture for their organization are identified and operationalized. Using conjoint analysis, the relative importance of these factors is measured as well as whether or not the effect of levels of these factors on decision-making is linear. The findings show that technology factors, which are a subset of classical diffusion variables, are sufficient to explain the adoption of computing architectures, which are a type of IS innovation with low impact on organizational processes and low knowledge barriers for end-users. The software quality associated with an architecture is the most important factor considered by IS managers and its effect is linear. The effect of the cost factor is less important, non-linear, and in some cases, unexpected. The effects of centralization, backward compatibility with the organization, and acceptance by third parties are all linear, but less important than software quality.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 43 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.