Fighting Misinformation on Social Media
[Music]
as you might have heard
or even experience yourself there is a
lot of worry these days
about misinformation that is being
shared on social media
these outlandish posts about 5g and
coronavirus
and hillary clinton endorsing donald
trump for example
were shared over a thousand times the
success of this kind of content
is both surprising and concerning
and has led into new fields of research
exploring how much misinformation is out
there
and what leads people to believe and
share it
but mostly attention has been paid to
the more important question
what can actually be done about this
problem
this is what i’ve been focusing on and
today
i’m going to tell you about one such
possible solution
in my search for solutions i started
asking the basic question
how why people share misinformation in
the first place
let’s look at these two headlines again
you don’t have to be a genius to tell
they are probably false
and yet each one was shared by thousands
of people
what were these people thinking or
actually
were they really even thinking at all it
is not actually obvious
what the impact of more thinking would
be on misinformation sharing
it could be that if people think more
they will be better able to identify
false claims
and therefore more discerning in what
they share
could also be that more thinking leads
to more motivated reasoning
leading people to seek out believe and
share
content that aligns with their ideology
even if it’s not true
teasing these two possibilities apart is
really important
as they offer opposite suggestions about
whether it will help
or hurt to get people to think more
so i set out to test whether analytical
thinking
helps people uncover truth or amplifies
bias
in my work i use the lens of cognitive
science
and consider people’s decision-making as
arising from two different modes of
information processing
they may stop and think carefully or
just go with their gut responses
and intuitions to measure the extent to
which
people rely on their analytical thinking
versus intuitive gut responses
i use the cognitive reflection test
which consists
of a set of questions with compelling
intuitive but drawing
answers for example if you’re running a
race
and pass the person in second place what
place are you in
the answer that intuitively comes to
mind for many people
is first place but think about it for a
second
this is not the correct answer if you
pass the person in second place
you are in second place the idea here is
to capture the extent to which
you just say the first thing that comes
into your head
versus stopping to think about it for a
second
in order to investigate the relationship
between users cognitive style
and quality of information they share on
social media
i designed a hybrid lab field study
first i had about 2 000 twitter users
fill out the cognitive reflection test
then i pulled information from their
twitter profiles
including the websites they shared
content from then to measure the quality
of information they shared
i used a list of 60 news websites whose
trustworthiness
had been rated by professional fact
checkers as part of a previous study
here are the results for each news
website
i showed the average cognitive
reflection tester score of users who
tweet links to that website on the
x-axis
and i ensure the website’s fact-checker
trustworthiness rating on the y-axis
the size of each dot is proportional to
the number of tweets to that
outlet in our data set as you can see i
found a significant
and positive relationship between
cognitive reflection
and information quality users who rely
more on analytical thinking
are more likely to share content from
websites that professional fact-checkers
rate as more trustworthy this results is
robust controlling for a variety of
demographic factors
including education income and political
partisanship
the pattern shown here suggests that
sharing misinformation
is more about failing to think carefully
rather than thinking
in a motivated way so what does this
suggest
about potential solutions to the
misinformation problem
if lack of careful thinking leads people
to share low quality content
then getting users to think about
accuracy before they share
could help them to be more discerning we
built an experiment to test this idea
and show how it could be done by social
media platforms
we messaged more than 5 000 twitter
users who had previously shared
low quality content we asked the users
to rate the accuracy
of a single non-political item we are
not telling them that
accuracy is important that they should
take accuracy into account
or any kind of normative statements like
that instead
you’re just priming the concept of
accuracy
making accuracy top of mind so when they
go back to their news feed
they’re more likely just on their own to
think about accuracy
as they decide what to subsequently
share
as you might expect almost no one
responded to the message
but as soon as they see they are treated
in that the concept of accuracy
has been activated in their mind we then
collected all tweets
from before and after receiving the
message and quantify the quality of
information using the same website’s
fact-checker transworthiness rating
as in the previous study here’s what we
found
for each news website i showed the
website’s factual curse throughout
ordinance rating
on the x-axis and i showed the change in
fraction of tweets containing links to
that website on the y-axis
so positive values mean the message
increase
the website’s share of tweets and
negative values mean the message
decrease the website’s share of tweets
as you can see
we find clear evidence that accuracy
message made users more discerning in
their subsequent sharing
after getting the message users shared
proportionate more links to high quality
mainstream websites
and fewer links to hyper partisan fake
news websites
as rated by professional fact checkers
does this mean that we should all
immediately get online and start yelling
at anyone we see sharing misinformation
not necessarily
in the experiment that i just told you
about we were very careful
to not make people defensive but in a
follow-up experiment we found people who
had shared
false content and publicly corrected
them by replying to their tweet with a
link to a fact checking website
unlike our subtle accuracy nudge this
direct correction
actually decreased the quality of
information they subsequently share
these studies suggest that people share
misinformation
because they forget to think about how
accurate the content is
before they share it the issue isn’t
necessarily that
people are lazy thinkers in general
instead
there is reason to believe that social
media by design distracts us from
accuracy
by focusing our attention on social
factors such as how many likes your
posts will get
what comments people will make and so on
so
even people who under normal
circumstances think very carefully
may wind up mistakingly sharing
misinformation while online
this also means that platforms can
change their design
social media companies have the power to
focus attention on accuracy
and help immunize against misinformation
if
they so choose