Emergencies in 140 characters

A TIHR Lunchtime talk concerning a current research project exploring the use and impact of social media in emergencies. The project is funded by the European Commission and involves partners in several European projects. Ultimately, the aim of the project is to improve the way emergency services make use of social media data – i.e. any information, photos or videos posted on social media such as Twitter or Facebook before, during or after an emergency.

Transcript

1. The use of social media by emergency services and ci5zens: ini5al findings from the FP7 EmerGent project Joe Cullen, Thomas Spielhofer, Kers5n Junge and David Drabble Tavistock Lunch5me Talk Series 29th July 2015
2. EmerGent: What is it? •  Research and Technology Development (RTD) project funded under the EC FP7 programme •  Explores use of social media in managing emergencies •  Aims to support beSer interac5on and informa5on exchange between emergency services and ci5zens •  Through guidelines, data mining and Apps Who is involved?
3. EmerGent structure and work plan
4. The Tavistock role in EmerGent •  TI mainly works in WP2 – ‘Impact of Social Media in emergencies’ (focusing on developing and tes5ng a concept and methodology for impacts assessment) •  And WP3 – ‘Analysis of social media in emergencies today and tomorrow’ (focusing on more technical aspects of use and impacts of social media in emergencies) •  Addi5onal role in WP7 – ‘Guidelines, Dissemina5on, Exploita5on and Ethics’ (responsible for se\ng up ‘Ethical Monitoring System’ and ensuring project and partner compliance with ethical requirements)
5. What WP2 does 5 T2.1 Concept T2.2 Impacts for EMS T2.4 EMS & Ci5zens involvement T2.3 Impacts for ci5zens T2.5 Expert Workshops Realist Review Case Studies Ac5on Learning D2.1: Concept D2.2: Impact for EMS & Ci5zens D2.3-2.6: Reports & workshops Theory of Change
6. 6 Emergenc y managem ent is inadequat e Ci5zens, EMS and social media are poorly connected Analysis of impact of social media Developing an App , data mining system and Guidelines Improved connec5vity Reduc5on in harm created by emergencies Basic Theory of Change for EmerGent
7. Challenges for TI •  What does impact mean? Are we inves5ga5ng the impact of social media on emergencies generally? The external impact of the project on individuals, organisa5ons, society? The impact of the workplan on Emergent internally? All three? •  No clear evidence base on impact of social media in emergencies •  Expected outputs and outcomes of EmerGent fluid and evolving – need for flexible, dynamic impact assessment framework •  Partners have different construc5ons of the project and its expected outcomes and impacts •  How to reflect the posi5ons and perspec5ves of people who do not use social media •  How to capture the social construc5on and adapta5on of EmerGent outputs and services – value-embedded ac5on systems •  Involving ci5zens and emergency services as ‘co-producers’ of knowledge
8. What we have done so far •  Ini5al concept for impact assessment •  Baseline theory of change •  Recent trends in social media usage •  Survey on emergency services and ci5zens use of social media •  Case study Toolkit •  Round 1 case study: The London Riots •  Set up Ethics Monitoring System •  Now at first stage of Round 2 case studies: ‘Flooding’
9. Social media – fact & figures If social media is used by emergency services: •  What propor5on of the popula5on are they talking with? •  How does this vary between European countries? •  Who are they talking to? •  Are these differences significant enough to affect the quality of social media data?
10. European social media figures •  Number of social media users: 329 million •  Propor5on of popula5on who use social media: 40 percent •  Propor5on of social media users who use mobile devices: 66 percent
16. CiGzen online survey •  1034 full responses (Feb-June 2015) •  About two-thirds (63%) aged 21-39 •  Roughly equal propor5on of males and females •  Most responses from Poland (30%), Slovenia (16%), Germany (16%), UK (14%) and Italy (7%) •  13% do not use a mobile/smartphone – 29% of 50+s •  19% work or volunteer for an ES and are significantly more likely to use SM etc. – ciGzen sample of 839 without these 16
17. Data Analysis •  Two factors: – Use of social media (7 items – α=.725) •  E.g.: ‘I regularly post messages on social media, such as Facebook, TwiSer or Instagram’; ‘Most of my friends use social media to keep in touch’ etc. – Value of using SM in emergencies (5 items – α=. 774) •  E.g.: ‘It is important for emergency services to use SM to keep in touch with the public during emergencies’; ‘Emergency Services should regularly monitor their SM so they can respond to such requests’ etc.. 17
18. Use of social media factor 18
19. Use of, and aOtudes towards using, social media (SM) in emergencies •  Young people use SM significantly more 19
20. Use of SM by age 20
21. Use of, and aOtudes towards using, social media (SM) in emergencies •  Young people use SM significantly more •  But there is no difference by age with regard to a\tudes towards using SM in emergencies •  Women use SM significantly more than men and also have a more posi5ve a\tudes towards use of SM in emergencies •  Ci5zens with children under the age of 18 are also more posi5ve towards using SM in emergencies (but no difference on normal use). 21
22. Using SM to find out info •  43% of ci5zens used SM to find out informaGon during an emergency and 58% thought it was likely they would do this in future (91% of those who had done so before vs. 32% of those who had not) •  Women are about 29% more likely than men to use SM to find out informa5on during an emergency •  Most used it to find out about: weather warnings (78%), road or traffic condi5ons (70%), damage caused by event (63%), or to see eyewitness photos or videos (60%) 22
23. Using SM to share info •  27% of ci5zens have previously used SM to share informaGon during an emergency – 48% thought they would do so in future •  Women are about 68% more likely to share informa5on via SM than men during an emergency (73 out of 373 men vs 145 out of 442 women) •  Most likely to share: weather warnings (66%), road or traffic condi5ons (64%) or an eyewitness photo (53%) •  Otherwise they used it to share: their feeling or emo5ons about what happened (41%), an eyewitness descrip5on (28%) or a video (22%) 23
24. Links to ci5zen survey •  ow.ly/KcED6 (English survey) •  bit.ly/1GnUUS2 (Link to surveys in other languages: Slovenian, Polish, German and Italian) 24
25. Round 1 case study: the London riots Background and raGonale •  August 2011 •  Part of a broader outbreak of disorder in England, sparked by death of Mark Duggan in ToSenham •  3.100 people arrested, 1000 charged •  186 police officers and 10 fire fighters injured •  Spreading of riots generally ‘blamed’ on social media, in par5cular BBM •  An example of a ‘social’ emergency created by humans and requiring a higher level of coordina5on than other emergencies •  Well documented •  1st of our mul5ple case study design
26. Social media in the London riots - BBM “If you’re down for making money, we’re about to go down hard in East London tonight, yes tonight!!! … Doesn’t ma=er if the police arrive ‘cos we’ll just chase dem out because as you’ve seen on the news they are NOT ON DIS TING. Everyone meet at 7 at StraKord Park and let’s get rich” “Everyone in edmonton enfield wood green everywhere in north linkup at enfield town staOon at 4 o clock sharp!”
27. Social media in the London riots - twiYer Rumors spread on TwiSer during the London riots People brought together through TwiSer gather to clear up Clapham Junc5on. (Taken by TwiSer user @Lawcol888) The London Eye on fire? Army vehicles deployed into the City of London?
28. Social media in the London riots - YouTube
29. Findings (1) – not all social media are similar ParGcular types of social media are more likely to be used by certain demographic and socio- economic profiles for parGcular purposes “The internet and that is a bit too bait, so no one really broadcasts it on the internet.(…) Like Twi=er there’s like a hashtag innit, like if someone hashtags riots you can go to that certain page and see what everyone has been saying about the riots. Police could easily go to that page there and see who’s been seSng up or organising groups to come”
30. Findings (2) – ci5zens’ roles and social media CiGzens take up different roles during emergencies –  Agitator’ / acGve parGcipant - “Everyone in edmonton enfield wood green everywhere in north linkup at enfield town staOon at 4 o clock sharp!” –  Informant: “someone … has just posted: Go on Hackney! Fuck the feds! #hackney Can we have them arrested for incitement pls?”’ –  Helper: “Everyday, gather us in an anO-riot. for this all you need to do is post a picture on this event with a nice cup of tea at home!” (#Opera5onCupofTea) –  Amplifier – someone who is able to spread a message widely and quickly because of the number of followers they have. –  CiGzen journalist – repor5ng and commen5ng on the events as they unfold –  Bystander – watch what is happening and pos5ng scenes on social networks
31. Findings (3) - rumours The mechanism of social media itself can funcGon as a correcGve mechanism for false rumors or speculaGon (1) A rumour starts with someone tweeting about the occurrence of an alleged incident. (2) The rumour gets retweeted (….). Some form of evidence – eyewitness reports, references to mainstream news sources, links to pictures (…) or to mainstream news sources on the Web, etc. – may be added as the original tweet gets retweeted and various reformulations of the rumour also begin to appear. (3) Others begin to challenge its credibility (i.e. make a counter-claim), perhaps on the basis of logical arguments (e.g. ‘it’s not possible because …’) or new information that throws into doubt the reliability of evidence previously offered. (4) A consensus begins to emerge (….). Where this is that the rumour is false, it may nevertheless re-surface in the corpus as latecomers pick up the original tweet and join in.
32. Findings (4) – Challenges for emergency services •  Scale of social media data –  Turning informa5on into intelligence is more challenging during extreme events resul5ng in millions of tweets and other social media data in a short period of 5me •  Organiza5onal and professional obstacles or facilitators to enhancing the use of social media data in emergencies –  Lack of technical equipment –  Staffing –  Procedures
33. London riots: immediate impact of social media Short term effects NegaGve PosiGve Spread informa5on about the event and influenced opinions about it Discouraging some people from par5cipa5ng in the riots Supported the spread of the riots beyond the immediate locality Organising ci5zens in cleaning up the damage caused by the riots Helped to coordinate the ac5vi5es of rioters. Raising money to compensate those nega5vely affected by the riots
34. London riots: longer-term effects of social media •  Changes in the policies and procedures of the MPS with regard to social media, e.g. –  Development of the @metpoliceuk TwiSer feed –  Crea5ng local borough level TwiSer accounts that are managed locally –  More followers •  A more sophis5cated engagement by the MPS at organisa5onal level –  Digital communica5ons steering group –  Designated MPS staff trained in social media engagement –  Improve the way the MPS engages with communi5es using social media
35. Next / current steps •  2nd wave case study: flooding •  Pan-European – Tes5ng and developing five key findings from London riots case using cases in UK, PL, D, GE •  Primary as well as secondary research •  Poten5ally some qualita5ve analysis of tweets