Future of Cloud-Based Services for Multi-factor Authentication: Results of a Delphi Study - Cloud Computing - page 137

RQ3: Success Factors (H20-H50). Since the success of AaaS solutions
directly depends on its adoption, we systematize possible success factors ac-
cording to the aforementioned adoption drivers. Furthermore, to enable deeper
insights, we differentiate success factor candidates at the different levels of an
AaaS solution. This includes the system implementation itself, one or more im-
plemented authentication methods , and organizational attributes specific to a
provider (ASP), offering at least one system. All hypothesized items are derived
from related literature [e.g., 3, 6] (see result tables 4, 5 & 6).
3.2 Applying the Delphi Method
The Delphi method can be defined as a structured group communication process
which allows individuals to deal with complex problems and has proven to be a
popular instrument in IS research and technology forecast [13, 16]. Here, classical
studies are characterized by the following attributes [10]: (1) Survey of selected
experts; (2) use of standardized questionnaires; (3) anonymity of individual re-
sponses; (4) calculation of statistical group answers; (5) iteration of the survey;
(6) provision of the group answers (controlled feedback) to the respondents.
The novelty, complexity and specificity of this paper's research object requires
the involvement of declared experts in related fields (e.g. Cloud Computing ).
Compared to alternative approaches like group discussions or expert surveys, the
Delphi method tends to reveal more reflected and thus better expert judgment
[10]. Major drawback, on the other hand side, is a higher expenditure of time
due to additional survey rounds conducted [10]. Essential for a high quality of a
Delphi study's generated results is the selection of experts with an appropriately
deep understanding of the research topic [10]. Related literature suggests a panel
size of 10-18 individuals or more which are selected non-randomly by the Delphi
monitoring team [10, 13, 16]. The panel should furthermore be composed inter-
disciplinary to cover a more faceted set of expert opinions [10].
Fig. 1. Process Model of the Study
The study follows the process depicted by figure 1. In a first step, the re-
search questions were specified, related contents systematized and a measure-
ment model derived. Afterward, potential experts were identified and selected
to join the expert panel. The expert panel was initially questioned in a non-
standardized form (Round (R) 1), and then in two successive standardized sur-
vey rounds (R2 and R3) with controlled feedback. After completion, the data
was analyzed and key findings were distributed to all active panel members.