Information Technology Reference
In-Depth Information
Fig. 4 (left)
Shows the frequency of response scaled to 1 for each status type (x-axis) and
excluding all self-ratings by participants. Intensity is described on the y-axis as ranging
from 0 (nonexistent) to 5 (very strong), and frequency is represented as a percentage on the
z-axis. (n = 14,051)
(right)
Shows a similar data categorization, but instead only including
responses where the participant has rated one of their own connections. (n = 2,934)
which participants considered as primarily “very weak.” RW indicates a connection
is only present in the real world. Such connections follow a more even distribution,
but one that remains heavily skewed towards weaker connections, although the
mean higher than FB. BOTH represents connections present in both spheres, which
are considered stronger than any other type of connection, but remain skewed
towards weaker connections than a normal distribution, supporting the Granovetter
[10] and Donath's [8] hypotheses. NOT represents connections that do not exist in
either the real or online worlds, and should have a frequency at intensity 0
approaching 1, as is observed. NBS represents connections not present on Facebook
or in the real world that the participant thinks should exist. These data conform to
the Granovetter [10] and Donath [8] hypothesis that users tend to have higher
numbers of connections that they consider weak.
In Figure 4(right), we see that users themselves have an increased bias towards
considering Facebook-only connections to be weak, and a proportionately higher
average intensity ranking for real world connections. Data in the BOTH category
shows little statistical difference but maintains the bias towards higher numbers of
weaker connections. This indicates that people perceive online connections to be
stronger than participants consider them.
Focusing at the level of the individual participant, there does appear to be a
spectrum in self-perceptions about connectivity strength. As Figure 5 shows, some
users consider their Facebook connections to be much more significant than
others, a result independent of friendship network size (not shown).
Figure 5 confirms Hypothesis (1), that many more weak connections exist than
strong on the group level, and indicates that, we can examine user accuracy as a
function of perceived strength. Hypothesis (2) states that connections perceived to
