Blurring the Boundaries? Ethical challenges in using social media for social science research
1. Blurring the Boundaries?
Ethical challenges in using social
media for social science research
Kandy Woodfield
Director of Learning
2. Blurring the boundaries?
Social media have blurred the boundaries, leading to
a set of linked challenges for researchers:
Methodological (volume, scale, nature of data)
Collaborative (trans/cross disciplinary teams)
Ethical & legal (privacy, ownership, control, power relations)
Capability (requires new skills, tools, frameworks,
infrastructure)
Contextual (understanding ‘the social world’)
Synthesis (how do new methods supplant, enhance, augment?)
3. New Social Media, New
Social Science?
Innovation
Collaboration
Inspiration
Fresh thinking
Network of methodological innovation
Funded by ESRC (via NCRM) initially
Now in its third year, self-funded, peer led,
network leads @
Affiliate organisations from academia, govt
and voluntary sector
4. Aims of the network
Innovation
Collaboration
Inspiration
Fresh thinking On & off line community of practice
Forge links across sectors & disciplines
Catalyse debate
Address challenges social media present for social science
research
Share approaches, tools & experiences of using social media
Identify good practice
Co-created content & guidance to be shared with the wider
community
5. How it works?
Range of platforms
Twitter: @NSMNSS, #NSMNSS
Blog: http://nsmnss.blogspot.co.uk/
You Tube:
www.youtube.com/user/NSMNSS
Conferences
Knowledge exchange
events
Methodological projects
Publishing
Innovation
Collaboration
Inspiration
Fresh thinking
6. 70% HE sector
30% other
60% in UK
40% worldwide
3,000+
Twitter community
20 disciplinary
fields
7. Over 18,000
minutes of
video watched
3 0nline seminars
2 conferences
7 knowledge exchange
seminars
17 twitter chats
140 blog posts
So far…
3,900+
video views
on
You Tube
160, 207blog page views
8. Ethical guidelines reviewed
Social media users & researcher’s
views explored
Crowd-sourced, book of blogs
9. What have we learnt about
social media for research?
Social media used in most soc sci disciplines
Lots of innovation but disciplinary silos &
divides do still exist and are counter-productive
to moving social media methodology(ies) forward
No single methodology for social media
research – many approaches, many tools,
different epistemological stances
Social media is a fast moving world, platforms,
data and users change = computational, ethical
and capability challenges
10. New social media, new social science and new ethical issues! Salmon, J. 2013
https://docs.google.com/file/d/0B1-gmLw9jo6fLTQ5X0oyeE1aRjQ/edi
What have we learnt? II
Persisting uncertainty about whether we are
‘getting it right’
Ethical dilemmas - lack of consistent, relevant
guidance, widely varying practices: ‘analogue ethics
in a digital age’
What are the big issues?
Avoiding the emperor’s new clothes
Understanding differences between aggregated
‘big data’ and qualitative socme data
Lack of research with users of social media
Lack of engagement with commercial platforms
11. Elizabeth A. Buchanan, Endowed Chair in Ethics, University of Wisconsin
Ethically unique?
Social media data is:
Malleable & ‘mashable’ – created by many,
invented and reinvented
Track-able
Mineable
Greasy - travels across platforms & borders
12. New social media, new social science and new ethical issues! Salmon, J. 2013
https://docs.google.com/file/d/0B1-gmLw9jo6fLTQ5X0oyeE1aRjQ/edi
Key ethical issues?
Privacy &
anonymity of
participants
Sampling &
recruitment
Informed consent
Researcher
identity, rapport &
relationships
User-generated
content (text &
images) as data
Date ownership
& data security
14. Qualitative research exploring:
How people curate their digital
lives
What they understand about how
their online information is used
What they think about their
information and posts being
used by researchers and in
online social media research
project?
Exploring social media users
views
Beninger, K. et al (2014)
http://www.natcen.ac.uk/media/282288/p0639-research-using-social-media-report-final-190214.pdf
15. Using social media
Used for leisure, social and
professional reasons
Familiarity with the
platform
Peer activity
Device mobility
Accessibility
Frequency of
use
16. Views about research Can/do users
distinguish between
academics using their
data & government or
commercial orgs?
18. Privacy in SM research
Problematic for researchers &
participants
Is the space seen as private by its users?
Are they aware they are being observed?
What is the researcher’s role?
Is everything what it seems?
Are users who they claim to be? Does it matter?
How do researchers ensure participants are anonymous?
IP addresses are (usually) traceable
Tweets may contain identifiers …
Twitter metadata contains geotags etc.
19. Informed consent
Terms & conditions of data use may require certain steps
Morally required even if T&Cs state it’s public/reusable?
What about bots/age/capacity? Difficult to verify w/out
informed consent
Promotes trust
Verify user views haven’t changed, been deleted, the ‘right
to be forgotten’
To publish photos or imagery
20. Terms of service
Who owns the data created in a social networking site?
Facebook claims the rights to any data collected from
applications (including surveys) created within it:
Jaquith (2009): “Facebook’s definition of data ownership
does not include the right to export that data. It’s “mine,” so
long as I leave it under Facebook’s control”
Twitter upholds the tweeters IP, insisting you use their
Twitter handle and verbatim when quoting in publications
21. What does this mean in
practice?
Twitter Best Practices for Media publication
Show name, @username, unmodified Tweet text
and the Twitter bird nearby, as well as a
timestamp
If displaying Tweets, make sure they are real, from
legitimate accounts and that you have permission
from the author when necessary
Real tension with research ethics & anonymity
20
22. Recruitment & Data Collection
Using data posted online
Digital identities & risks for users
Who has a right to privacy online?
Exclusion of particular groups – whose voice, whose
stories?
Even data pools are made up of human individuals –
with geo tags and strong views, distinctive
characteristics revivification of identity can be possible
Just because it’s possible doesn’t mean it’s ethical
24. What does this mean in practice?
Collecting qualitative data from users on social
media:
Pull rather than push recruitment – attract to you, caveat -
calls for participation can get modified, amended
Approach gatekeepers of communities
Be transparent and open about your researcher role (if
doing ethnography, observation)
Think about informed consent process –conversations
online draw in other users inadvertently what about their
consent
Informed consent needs to be a process, not a one off
23
25. Researcher identity &
wellbeing
Not mentioned by users but researchers
were concerned about…
Their own digital identity & footprint
Impact on research outcomes
Managing communication w.
participants – safety and disclosure
Credibility and transparency
26. Analysis & presentation of data
Analysis
Third Party Software – who controls the sample, the
feed, owns the data
How much is too much? Perils of network analysis…
Validity and representativeness
Presentation
Traceability of participant data - anonymity
short & long term implications for participants
Terms of usage – Twitter and Facebook controls on how to
display data and posts
Journals differ in their approach to use of verbatim posts
27. Researchers
Is SM the right methodology for your research Q?
Don’t make assumptions
Review case studies and existing research
Recruitment:
• Transparency in materials
• Learn about privacy terms of the platform used
Collecting/generating data:
• Considering implications of legally permitted vs. intellectual property
• Acknowledge limits of accessing different user types, population groups
Reporting results:
• Test traceability of data, and paraphrase or remove handle
• Reasonably seek consent for use of verbatim/sensitive content
28. What have we learnt? III
‘Getting it right’ is also about
methodological quality:
What is a robust sample from Twitter or
Facebook?
Need to develop methodological courage and
confidence to defend the method
Need new quality frameworks?
Scepticism and cynicism persist
Digital literacy & methodological skills gaps
Lack of experience and understanding in
institutions, ethics boards and funders
29. Resolving ethical
dilemmas requires
“holistic” approach
engaging views of many
disciplines
Cultural competencies &
flexibility needed when
dealing with multicultural
participants
Need for ‘standardization’
of REC approaches?
Social media are
plural, no single
methods – ethics
decisions must
be context and
method specific
Encouragement
for researchers to
publish methods/
ethical case
studies, failures &
successes
30. Some final observations…
Multi-disciplinary, multi-method approaches to research
ethics are needed, while respecting the influence of disciplinary
codes
Ongoing development & dialogue is needed approaches will
evolve with changes in tech & usage – static codes won’t work
Must engage those who teach, review & supervise
Continue to support discussion of ethical issues / case
studies
Need to start a dialogue about ethics with the software
houses
We’re not there yet
31. Thank you!
If you want further information or would like to
contact the network:
nsmnss@natcen.ac.uk
http://nsmnss.blogspot.co.uk/
@nsmnss on Twitter
Kandy can be contacted via LinkedIn or @jess1ecat
Notes de l'éditeur
Hello and thanks for coming to this session.
Intro self and NatCen
Explain speak for about half hour then open up to discussion/Q’s for 15 minutes
Encouraged to tweet and use your phone
What you are all here for is to discuss some of the challenges associated with social media research.
As an ever-optimist, I like this think of these as opportunities for innovation, and analytical and creative problem solving rather than challenges! Either way, I am here to flag up some key challenges faced by practitioners.
Not an exhaustive list, and some of these will appear more or less so in different methodologies
What you are all here for is to discuss some of the challenges associated with social media research.
As an ever-optimist, I like this think of these as opportunities for innovation, and analytical and creative problem solving rather than challenges! Either way, I am here to flag up some key challenges faced by practitioners.
Not an exhaustive list, and some of these will appear more or less so in different methodologies
What you are all here for is to discuss some of the challenges associated with social media research.
As an ever-optimist, I like this think of these as opportunities for innovation, and analytical and creative problem solving rather than challenges! Either way, I am here to flag up some key challenges faced by practitioners.
Not an exhaustive list, and some of these will appear more or less so in different methodologies
What you are all here for is to discuss some of the challenges associated with social media research.
As an ever-optimist, I like this think of these as opportunities for innovation, and analytical and creative problem solving rather than challenges! Either way, I am here to flag up some key challenges faced by practitioners.
Not an exhaustive list, and some of these will appear more or less so in different methodologies
Recruitment
Participants are sceptical! (this theme permeates the presentations)
[Scepticism: confused about how researchers would use social media posts. Lack of understanding of the value of social media research. Sceptical about quality of data you would get because of a range of issues such as exaggerated views, sampling limits; Acceptance: reduce interviewer effect as not happening face to face. Trends; data mining for example could get enough views to be acceptable; could give understanding to a social issue; Ambivalence: some participants held a fatalistic view- regardless of what they think researchers will use their information]
Will a growth in user scepticism abt data usage by agencies, companies mean more personal encryption & narrowing/biasing of data?
Digital identities and pseudonyms: how to verify who your participants are; Tension between online and offline personas.
Issue of what is a person and what isn't? IE. Is an avatar a person? Is your digital info an extension of yourself and therefore should be treated as you would a person in research?
Exclusion of particular groups: without online access or no access at home- who is in your sample? How representative is it of your population target?
Informed Consent
While views were not unanimous about gaining informed consent and protecting anonymity, the view that some action should be taken was very prominent.
Reasons for not doing so are: (BOTH RESEARCHERS AND USERS HOLD THESE VIEWS)
user’s responsibility
site’s responsibility- site owners should make it more clear
Reasons for informed consent:
Morally and legally required
Promote trust
Verify user views haven’t changed
To publish photos or imagery
Anonymity needed, especially if informed consent not gained, to avoid harm such as judgment and ridicule, to preserve reputations
Data Collection
Ownership: legal-country + site laws; intellectual property
User expectations vs. researchers’ vs. the law
Users probably didn’t expect their blog or their tweets to be used in research. Must consider their expectations at the time of using a site. Varies by site ie. Membership forum they felt very strongly its private and tends to be used for more sensitive discussions whereas twitter’s set up means its public.
Researchers divided as well- some think that if you put it out there its public, how does it differ from observational research?
Law varies. For example, case in northern ireland last year in which private messaging in fb was admissible in court.
Expectations linked to the type of site- those deemed more private or public than others.
Huge variation in types of sites and forms of interaction on each (text, live or recorded audio, images or video): each has different and evolving issues
Public vs. private sites: ambiguous, contested, changing
Twitter vs membership only forum
No hard and fast rule- each site has different t&cs , might be constrained by laws of that country
Recruitment
Participants are sceptical! (this theme permeates the presentations)
[Scepticism: confused about how researchers would use social media posts. Lack of understanding of the value of social media research. Sceptical about quality of data you would get because of a range of issues such as exaggerated views, sampling limits; Acceptance: reduce interviewer effect as not happening face to face. Trends; data mining for example could get enough views to be acceptable; could give understanding to a social issue; Ambivalence: some participants held a fatalistic view- regardless of what they think researchers will use their information]
Will a growth in user scepticism abt data usage by agencies, companies mean more personal encryption & narrowing/biasing of data?
Digital identities and pseudonyms: how to verify who your participants are; Tension between online and offline personas.
Issue of what is a person and what isn't? IE. Is an avatar a person? Is your digital info an extension of yourself and therefore should be treated as you would a person in research?
Exclusion of particular groups: without online access or no access at home- who is in your sample? How representative is it of your population target?
Informed Consent
While views were not unanimous about gaining informed consent and protecting anonymity, the view that some action should be taken was very prominent.
Reasons for not doing so are: (BOTH RESEARCHERS AND USERS HOLD THESE VIEWS)
user’s responsibility
site’s responsibility- site owners should make it more clear
Reasons for informed consent:
Morally and legally required
Promote trust
Verify user views haven’t changed
To publish photos or imagery
Anonymity needed, especially if informed consent not gained, to avoid harm such as judgment and ridicule, to preserve reputations
Data Collection
Ownership: legal-country + site laws; intellectual property
User expectations vs. researchers’ vs. the law
Users probably didn’t expect their blog or their tweets to be used in research. Must consider their expectations at the time of using a site. Varies by site ie. Membership forum they felt very strongly its private and tends to be used for more sensitive discussions whereas twitter’s set up means its public.
Researchers divided as well- some think that if you put it out there its public, how does it differ from observational research?
Law varies. For example, case in northern ireland last year in which private messaging in fb was admissible in court.
Expectations linked to the type of site- those deemed more private or public than others.
Huge variation in types of sites and forms of interaction on each (text, live or recorded audio, images or video): each has different and evolving issues
Public vs. private sites: ambiguous, contested, changing
Twitter vs membership only forum
No hard and fast rule- each site has different t&cs , might be constrained by laws of that country
Recruitment
Participants are sceptical! (this theme permeates the presentations)
[Scepticism: confused about how researchers would use social media posts. Lack of understanding of the value of social media research. Sceptical about quality of data you would get because of a range of issues such as exaggerated views, sampling limits; Acceptance: reduce interviewer effect as not happening face to face. Trends; data mining for example could get enough views to be acceptable; could give understanding to a social issue; Ambivalence: some participants held a fatalistic view- regardless of what they think researchers will use their information]
Will a growth in user scepticism abt data usage by agencies, companies mean more personal encryption & narrowing/biasing of data?
Digital identities and pseudonyms: how to verify who your participants are; Tension between online and offline personas.
Issue of what is a person and what isn't? IE. Is an avatar a person? Is your digital info an extension of yourself and therefore should be treated as you would a person in research?
Exclusion of particular groups: without online access or no access at home- who is in your sample? How representative is it of your population target?
Informed Consent
While views were not unanimous about gaining informed consent and protecting anonymity, the view that some action should be taken was very prominent.
Reasons for not doing so are: (BOTH RESEARCHERS AND USERS HOLD THESE VIEWS)
user’s responsibility
site’s responsibility- site owners should make it more clear
Reasons for informed consent:
Morally and legally required
Promote trust
Verify user views haven’t changed
To publish photos or imagery
Anonymity needed, especially if informed consent not gained, to avoid harm such as judgment and ridicule, to preserve reputations
Data Collection
Ownership: legal-country + site laws; intellectual property
User expectations vs. researchers’ vs. the law
Users probably didn’t expect their blog or their tweets to be used in research. Must consider their expectations at the time of using a site. Varies by site ie. Membership forum they felt very strongly its private and tends to be used for more sensitive discussions whereas twitter’s set up means its public.
Researchers divided as well- some think that if you put it out there its public, how does it differ from observational research?
Law varies. For example, case in northern ireland last year in which private messaging in fb was admissible in court.
Expectations linked to the type of site- those deemed more private or public than others.
Huge variation in types of sites and forms of interaction on each (text, live or recorded audio, images or video): each has different and evolving issues
Public vs. private sites: ambiguous, contested, changing
Twitter vs membership only forum
No hard and fast rule- each site has different t&cs , might be constrained by laws of that country
Emperors new clothes: same issues as ever?
Issue of human subjects – what about data generated by bots, avatars etc – focus instead of harm, vulnerability, personally identifiable data etc.
Data/text/person - ,every data pool begins with a person
Anonymity (for example SNS posts) - quotes can help build authenticity but risk harm – on SNS platforms interactions may be public, check settings use DMs and emails instead?
When and what will you be using of people’s posts, - demonstrate relevance and limits of data collection
Comments by others add depth and richness to accounts but not usable? Or are they?
Who owns that data?
Remind people – informed consent as an ongoing process (inhibitions drop)
Public/private changing and contested public, individual conceptions of privacy and expectations are fluid (see our research) Data aggregators make data accessible to far wider audience than thought
Digital identity-
How much to share? Disclosing identities as researchers, and how
being consistent and keeping professional identity separate from personal;
Observing boundaries between personal and professional identities online;
using your online identity to gain trust and credibility from potential participants
safety
Impact on research outcomes= Known presence influence behaviours?
Managing communication w participants:
-In traditional research there are boundaries, and expected norms when the participant would/wouldn’t give you their views. With online research, esp. if all contact with the participant is online, then there is no limit to the level of contact/how often/when the participant is telling you their views.EX: One of the researchers was saying she really struggled with this and it really affected her wellbeing, as she was constantly having to deal with participants at all hours of day and night. There was no way for her to 'switch-off' from the constant flow of information, and she found it really hard to deal with.
Analysis:
A reoccurring theme amongst researchers, and indeed, strongly raised by SM users was of topics related to the ethical principles of validity and representativeness.
People behave differently online and offline and so online research could not reflect the ‘real world’
Exaggerated views were a result of the anonymity the internet afforded and therefore research findings using views from online sources would lead to inaccurate conclusions about something or someone.
Impulsive comments posted online may result in researchers using a view that does not accurately reflect someone’s ‘normal’ viewpoint but instead only something they held for a moment in time
Inaccurate profiles taken without further context would lead to inaccurate information and findings.
THIRD PARTY SOFTWARE: Using third party software and understanding what that software does – so if you don’t understand how it’s collecting data or how it’s presenting it through visualisations, how can you be confident about your analysis? Given that in other areas of research we know how we sample, we know how we develop a topic guide and we know how we did the analysis, how can we be confident about an analysis that we may not completely understand?
I.E NVivo's twitter import tool doesn't even technically let you anonymise what you import
How much data do you really need to do your analysis – because it’s available and because there’s so much, there’s a temptation to just take everything, but I think we veered towards saying that you should define what it is you need, and stop there, rather than just taking everything because it’s out there.
Presentation of data:
Traceability of tweets and other online content- most contain enough personal info and metadata that may enable users to be identified. More accessible and permanent than f2f encounter like an interview. HUGE concern of our users bc of concerns like professional reputation, personal safety, ridicule
short & long term implications for participants abt how you use their data