Cryptography Reference
In-Depth Information
two measures were usually negatively correlated with each other (Pearson correlation
coefficient was found to be -0.771, when considering data from all users), in certain
cases the correlation was not entirely clear. In order to understand an overall impact of
Toughness and Satisfaction on the usability of PMs, we pursued principle component
(PCA) and cluster analysis. Due to ease of readability, we do not include the details
regarding this analysis (a similar analysis, however, is later discussed for Post Test
measures in Section 5.2). We only depict the results (using Agglomerative Hierarchical
Clustering) of this analysis in Figure 3.
For All users, we obtain that Online
USB
Phone, and USB and Phone are
clustered together (we use '
' to denote preference). A similar and independent PCA
and cluster analysis for Students and Non-Students indicate the following. For students,
Online
Phone, and Online and USB form a cluster of their own. On the other
hand, for Non-Students, Online
USB
Phone, and USB and Phone are clustered
with each other. These results are intuitive and very much inline with our observations
shown in Figure 2, which we discussed in the previous subsection.
USB
3.5
3.5
3.5
3
3
3
2.5
2.5
2.5
2
2
2
1.5
1.5
1.5
1
1
1
0.5
0.5
0.5
0
0
0
(a) All Users
(b) Students
(c) Non-Students
Fig. 3.
During Test Cluster Analysis Based On Principal Components (Dissimilarity vs. PM)
Usability of Individual Tasks:
As we explained in Section 4.2, in our experiments,
each PM was tested for several different processes, namely, Register, Login, Second
Login, Change Password, Login with New Password. Since usability of a PM depends
on all these processes, we compare the three PMs based across these processes.
Figure 4 depicts the average Likert scale Toughness ratings for different processes
corresponding to Students, Non-Students and All Users. In this plot, first three bars for
each process correspond to Students (USB, Phone, Online, resp.), next three bars cor-
respond to Non-Students (USB, Phone, Online, resp.) and last three bars correspond to
All Users (USB, Phone, Online, resp.). The Satisfaction ratings were generally inversely
related to the Toughness ratings and are not shown in this paper.
Let us first compare the three PMs across different processes. We note that for each
process, in general, Phone is tougher than the other two PMs. Between USB and Online,
the former is deemed tougher, for all processes. This analysis conforms well with our
overall analysis of During Test data presented in previous subsections.
There are a few exceptions to the above claim, however. Login and Change Password
have the same average ratings for both USB and Online for Students Students deemed
Login with New Password as equally tough for USB and Phone. Register was also rated
at a equal level of toughness by Non-Students. For Second Login, Students found USB
Search WWH ::
Custom Search