[Air-L] [Request for Input] Research Practices for Closed Messaging Groups
gvrkirann at gmail.com
Wed Mar 25 14:19:14 PDT 2020
I am a postdoc at MIT and one of my primary projects here is based on
analyzing large amount of data we got from public groups discussing
politics in India.
We do not have any work published yet, but I've been in this space for over
a year now and have been thinking about similar questions.
I was the authors of one of the papers to developed tools to collect data
from public whatsapp groups.
There we defined public groups as groups which had a link to join publicly
During the recent elections in India, we used the same methodology to
collect data from thousands of public groups discussing politics. We did
not explicitly declare our presence in these groups. However, our WhatsApp
account clearly mentioned that we are a research institution and collecting
this data for research. We have IRB from MIT for our research. We take
extra care in storing personally identifiable information. We store the
anonmyized message data separately from the personal information and do not
share any information unless aggregated.
Similar attempts to monitor public groups have been done elsewhere, e.g. at
Oxford: https://comprop.oii.ox.ac.uk/research/india-election-memo/ where
they explicitly declared their presence in the group once they joined them.
they did not monitor groups with less than 60 users.
There have been other works in Brazil
http://www.dcc.ufmg.br/~fabricio/download/resende-www2019.pdf but I do not
think the authors declared their presence.
Other work I am aware of is from people at Meedan (Scott Hale) and
FirstDraft (Claire Wardle) who have been trying to use data from fact
checking tiplines on WhatsApp that were set up to collect data during the
elections in Brazil and India. This is a different model to the 'public'
data collection above as it is opt in.
On Wed, Mar 25, 2020 at 11:35 AM connie im dialog <connieimdialog at gmail.com>
> Dear all,
> Hope everyone and your loved ones are doing well.
> As part of a project connected to MisinfoCon <http://misinfocon.com/> and
> Carter Center <http://cartercenter.org/>, and also in conversation with
> Secretariat at the Forum on Information and Democracy
> I have been overseeing a landscape review of current practices related to
> researchers entering into closed messaging spaces (e.g. WhatsApp,
> Telegram). This landscape review is intended to inform a set of
> recommended practices related to closed messaging spaces to be developed in
> conversation among human rights practitioners, election observers, and
> fact-checking projects.
> We are seeking protocols and examples of research practices mentioned in
> papers, in addition to how these efforts define privacy and public
> discourse; the particular backdrop has been elections but can be broader.
> Examples of issues that we are focusing on:
> * how the study defines *public* and *private*
> * whether the size of the group matters and how
> * whether and how the researchers identified themselves
> * what protocols have been defined related to de-identification of
> participants, limitation of content collected, and the storage around the
> * whether the study protocol has been reviewed by an IRB or other external
> Should you have recommendations of papers and reports to include, I would
> love to hear about them here or off-thread. If you are interested also in
> participating in the effort overall, please feel free to reach me
> off-thread as well.
> connie moon sehat
> connieimdialog at gmail.com
> PGP Key ID: 0x95DFB60E
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> Join the Association of Internet Researchers:
More information about the Air-L