Using Twitter as a data source: An overview of ethical challenges
Download as PPT, PDF0 likes391 views
The document provides an overview of ethical challenges encountered in using Twitter data for social media research, particularly during infectious disease outbreaks by a PhD student at the University of Sheffield. It discusses ethical issues such as informed consent, data confidentiality, and the implications of analyzing public data, alongside guidelines for completing an ethics application. The presentation also addresses recommendations for improving research practices and the importance of ethical considerations in social media research.
Using Twitter as a data source: An overview of ethical challenges
- 1. Using Twitter asadata source: An overview of ethical challenges Wasim Ahmed (wahmed1@sheffield.ac.uk) Prof Peter Bath (p.a.bath@sheffield.ac.uk) Dr Gianluca Demartini (g.demartini@sheffield.ac.uk) Ethics and Social Media Research Conference Monday 21st of March, 2016
- 2. About me • Second Year PhD student in the Health Informatics Research Group, Information School, University of Sheffield • PhD examines content that is shared on Twitter during infectious disease outbreaks • Run a social media research blog (over 9,500 hits) • Twitter Manager for NatCen Social Research’s New Social Media New Social Media (#NSMNSS) network.
- 3. Overview of presentation • Part 1: Background to ethical issues encountered (as a blogger and within PhD) • Part 2: Completing an ethics application for purposes of PhD • Part 3: Ethical issues and challenges that are discussed within the NSMNSS network
- 4. Part 1 : Ethical IssuesEncountered • PhD topic uses Twitter as a primary source of data – no interviews/surveys • Non-traditional data for a social science PhD: concerns over informed consent, and data confidentiality • Due to the volume of tweets it is impossible to obtain informed consent from all Twitter users
- 5. Part 1 : Ethical IssuesEncountered • Synthesized and blogged about software to retrieve data from Twitter. Twitter data in less than 5 minutes • Main critique of post –ethical implications of using Twitter tools? • Software developers: data is the in public domain? But is this the case?
- 6. Part 2 : EthicsApplication • Do I need ethics approval: Twitter data is publically available for everyone to see? • Ethics approval required as when the data is analysed things may emerge from the data that could draw attention to groups, individuals, trends etc. • Beyond what would normally be expected from engagement on these platforms
- 7. Part 2 : EthicsApplication • Ethics application covered: • Potential participants: who are the Twitter users being analysed: general public, organisations, public figures, specific geographical locations, or all? • For Twitter, participants are those whom use specific keywords & hashtags. • Data confidentiality and data storage measures: data is stored on secure laptops.
- 8. Part 2 : EthicsApplication • Methods used may have ethical implications e.g., crowdsourcing (used in some articles) • My ethics application listed: semantic analysis, sentiment analysis, and thematic analysis • Amendments can be requested if a new method of analysis is to be used to analyse Twitter data
- 9. Part 2 : EthicsApplication • Consent: • Ethics application made it possible to gain consent to use tweets / user handles within publications • Also possible to gain consent via a tweet if it was not possible for participants to view a participant information sheet and complete a consent form • Electronic versions of participant information and consent sheets, for example, via a Google form.
- 10. Part 2 : EthicsApplication • Majority of analysis is aggregate – identifying themes / clusters of tweets • Individual tweets and/ or user handles in original form will never be published • Data is analysed confidentially and never in public
- 11. Part 3: Ethical issuesdiscussed within NSMNSSnetwork • Public vs. private/ Facebook vs. Twitter – is the space being researched seen as private by its users? • Twitter datasets may include data generated by minors – sometimes overlooked. • Very difficult to filter Twitter data for under 18s
- 12. Part 3: Ethical issuesdiscussed within NSMNSSnetwork • Although not of direct relation to my PhD – there are cases of ‘suspect’ social media research in the media • Important to understand as to avoid potential pitfalls
- 13. Casestudiesdiscussed in NSMNSSnetwork • Samaritans Radar app designed to detect when people on Twitter appeared to be suicidal – used an algorithm to identify key words and phrases which indicated distress. • Users who have signed up for the scheme would receive an email alert if someone they followed tweeted such statements.
- 14. Casestudiesdiscussed in NSMNSSnetwork • Timeline of events • 30th October 2014 - reassure users and mention Whitelist • 31st October – mention testing and research input • 2nd of November – Mention subscribers (3 thousand), and 20 thousand Twitter mentions and trending for2 days. • 4th November – Offer reassurance and that they have listened to feedback • 7th November – Apologise for any distress caused to the public due to the range of information and opinion on the app. Suspend the app. • 14 November – Further apologise for any stress offer number, email, and website • 10th of March 2015 – Confirm that the app has been permanently deleted Launched 29 October 2014 and suspended on 7th November 2014
- 15. Casestudiesdiscussed within NSMNSSnetwork • Facebook emotion study (Jan 2012) positive posts from 155,000 Facebook users were removed • Issues of using scrapped data sets e.g., Reddit dataset containing every publically available Reddit comment. • Ted Cruz using firm that harvested data on millions of unwitting Facebook users
- 16. What next? • No fast answers but some good work already: Report by NatCen - Research using Social Media; Users’ Views • Findings: Participant’s views about research using social media fell into three categories: scepticism, acceptance and ambiguity.
- 17. What next? • Research using Social Media; Users’ Views Suggestions for improving research practices: •Sampling and recruitment – be transparent in purpose and aims in order to ethically recruit participants to online and social media research
- 18. What next? • Research using Social Media; Users’ Views Suggestions for improving research practices: •Collecting or generating data – to improve representativeness of findings and to understand privacy risks of platform used in a study in order to uphold protection
- 19. What next? • Research using Social Media; Users’ Views Suggestions for improving research practices: •Reporting results – to protect the identify and reputation of participants, maintain their trust in the value of the research and contribute to the progression of the field by being open and honest in reporting.
- 20. • Wisdom of the Crowd ‘#SocialEthics’ report (Ipsos MORI and Demos/CASM) • Findings: public awareness that information on social media can be mined for research is low compared to other uses of social media data What next?
- 21. Recommendations for Researchers: •Researchers to work with developers – only collect data that is required by the project, and remove if not required •Move to a culture of questioning whether the data being collected is really necessary for a research project What next?
- 22. • Examples of data minimization for a project may include: • Removing author’s name and @tag/userhandle from researchers sight • Stripping out other data that is downloaded e.g., named persons or place names What next?
- 23. • Examples of data minimization for a project may include: • Removing metadata that is not relevant for the purposes of a research project such as GPS data that might be attached to the social media post • Creating generalized groupings of data rather than analysing specific data e.g., by cities instead of exact street locations What next?
- 24. Conclusion • Genuine ethical issues around researching social media • Academic researchers refer to best practice guidelines when conducting social media research • Ethics application allows you to think through ethical implications from the beginning of the research process 03/22/16 © The University of Sheffield