it. Underestimating the audience might also be a comfortable equilibrium for some users who feel more comfortable
speaking to a relatively small group and would not post if they
knew that they were performing in front of a large audience.
On the other hand, many users expressed a desire for a larger
audience, and demonstrating that they do in fact have a large
audience might make them more excited to participate.
Pragmatically, this work suggests that social media systems
might do well to let their users know that they are impacting
their audience. Because so many members of the audience
provide no feedback, there may be other ways to emphasize
that users have an engaged audience. This might involve emphasizing cumulative feedback (e.g., 400 friends saw their
content last month), showing relative audience sizes for posts
without sharing raw numbers, or encouraging alternate modes
of feedback [3].
CONCLUSION
We have demonstrated that users’ perceptions of their audience size in social media do not match reality. By combining
survey and log analysis, we quantified the difference between
users’ estimated audience and their actual reach. Users underestimate their audience on specific posts by a factor of
four, and their audience in general by a factor of three. Half
of users want to reach larger audiences, but they are already
reaching much larger audiences than they think. Log analysis of updates from 220,000 Facebook users suggests that
feedback, friend count, and past audience size are all highly
variable predictors of audience size, so it would be difficult
for a user to predict their audience size reliably. Put simply,
users do not receive enough feedback to be aware of their
audience size. However, Facebook users do manage to reach
35% of their friends with each post and 61% of their friends
over the course of a month.
Where previous work focuses on publicly visible signals such
as reshares and diffusion processes, this research suggests
that traditionally invisible behavioral signals may be just as
important to understanding social media. Future work will
further elaborate these ideas. For example, audience composition is an important element of media performance [27],
and we do not yet know whether users accurately estimate
which individuals are likely to see a post. We can also isolate
the causal impact of estimated audience size on behavior, for
example whether differences in perceived audience size cause
users to share more or less. Finally, these results suggest that
there are deeper biases and heuristics active when users estimate quantities about social networks, and these biases warrant deeper investigation.
ACKNOWLEDGMENTS
We would like to thank our survey participants for their insight and time. In addition, we thank Cameron Marlow, Dan
Rubinstein, Sriya Santhanam and the Facebook Data Science
Team for valuable feedback.
it. Underestimating the audience might also be a comfortable equilibrium for some users who feel more comfortablespeaking to a relatively small group and would not post if theyknew that they were performing in front of a large audience.On the other hand, many users expressed a desire for a largeraudience, and demonstrating that they do in fact have a largeaudience might make them more excited to participate.Pragmatically, this work suggests that social media systemsmight do well to let their users know that they are impactingtheir audience. Because so many members of the audienceprovide no feedback, there may be other ways to emphasizethat users have an engaged audience. This might involve emphasizing cumulative feedback (e.g., 400 friends saw theircontent last month), showing relative audience sizes for postswithout sharing raw numbers, or encouraging alternate modesof feedback [3].CONCLUSIONWe have demonstrated that users’ perceptions of their audience size in social media do not match reality. By combiningsurvey and log analysis, we quantified the difference betweenusers’ estimated audience and their actual reach. Users underestimate their audience on specific posts by a factor offour, and their audience in general by a factor of three. Halfof users want to reach larger audiences, but they are alreadyreaching much larger audiences than they think. Log analysis of updates from 220,000 Facebook users suggests thatfeedback, friend count, and past audience size are all highlyvariable predictors of audience size, so it would be difficultfor a user to predict their audience size reliably. Put simply,users do not receive enough feedback to be aware of theiraudience size. However, Facebook users do manage to reach35% of their friends with each post and 61% of their friendsover the course of a month.Where previous work focuses on publicly visible signals suchas reshares and diffusion processes, this research suggeststhat traditionally invisible behavioral signals may be just asimportant to understanding social media. Future work willfurther elaborate these ideas. For example, audience composition is an important element of media performance [27],and we do not yet know whether users accurately estimatewhich individuals are likely to see a post. We can also isolatethe causal impact of estimated audience size on behavior, forexample whether differences in perceived audience size causeusers to share more or less. Finally, these results suggest thatthere are deeper biases and heuristics active when users estimate quantities about social networks, and these biases warrant deeper investigation.ACKNOWLEDGMENTSWe would like to thank our survey participants for their insight and time. In addition, we thank Cameron Marlow, DanRubinstein, Sriya Santhanam and the Facebook Data ScienceTeam for valuable feedback.
การแปล กรุณารอสักครู่..
