Presentation on theme: "Inoculation Strategies for Risk Communication Messaging & Modeling the Dynamics of Risk Perception and Fear: Examining Amplifying Mechanisms and Their."— Presentation transcript:
Inoculation Strategies for Risk Communication Messaging & Modeling the Dynamics of Risk Perception and Fear: Examining Amplifying Mechanisms and Their Consequences Using Twitter and Survey Data Represents a collaboration between research groups at University of Kentucky and Decision Research January 5 2015 Jeannette Sutton-University of Kentucky (PI and TL) William J. Burns-Decision Research (PI and TL) Paul Slovic-Decision Research (Co-PI)
2 Project Objectives: Research Goals To test for presence of affective/emotion in response to terrorism risk signals in a large dataset of Tweets Compare results from Twitter and survey data of the same event and time period Research Transition Goals To produce a report and publication on our findings and to recommend approach to combining results from Twitter and survey data that helps TSA under stand public reaction to TSA policy. Recommend approach to combining results from Twitter and survey data that helps TSA understand public reaction to TSA policies and communications.
3 DHS Interest and Motivation: Relevance to DHS DHS is interested in public uses of social media and terse messaging (Twitter) following a terrorist event. TSA has used twitter data to inform their marketing and risk communication. DHS is interested in learning about the impacts of terrorism on public risk perception, which can be observed in twitter data in response to terrorist events The Paperwork Reduction Act makes it very difficult for the TSA to conduct public surveys and so they need to understand to make the best use of publicly available data (e.g. Twitter data) DHS Contacts Ken Fletcher TSA Jerry Booker TSA Jerry Koehler TSA David Lim TSA Todd Trafford TSA
4 Potential non-DHS Stakeholders: Who else (operators/customers) could be interested in your research transition? –National Counterterrorism Center –LA and San Diego Fire Departments
5 Interfaces to Related Research Interfaces with others in this field Carter Butts (University of California, Irvine) – Network analysis and modeling Eduard Hovy (Language Technology Institute at Carnegie Mellon University) – Maritime risks Robin Dillon-Merrill (Georgetown University) - Near misses Richard John and Heather Rosoff (USC) – Perceptions of risk and emotional response to disasters Tim Sellnow (University of Kentucky) – Inoculating risk communication
6 Research Technical Plan The overarching plan is to compare public response to the Boston attacks using Twitter and survey data from a national panel. RQ1: what affect pulses can be identified in response to terrorism risk signals longitudinally? RQ2: what emotion is relayed in response to terrorism risk signals? RQ3: what is the relationship between terrorism risk signals and affect pulses longitudinally? RQ4: how does the emotional content expressed spontaneously in Twitter data compare with that expressed in survey data?
7 Research Technical Plan:RQ1 Dataset: 3/15/13 - 9/15/13 “terrorism” = 1.15 million tweets “terrorism + boston” = 46,000 tweets Dataset provided by the HEROIC project (Sutton and Butts, Co-PIs) funded by NSF-CMMI- 1031779)
10 National Survey Data Data that tracked a national panel’s response to the Boston attack will be analyzed in terms of level of different emotions and how they changed over time. Please see example table below. Public Response April 16 2013April 30 2013July 19 2013 AngerMean level, percent registering high Fear Sadness Perceived Risk
11 Research Technical Plan: Next steps… Qualitatively code a random sample of tweets from risk signal windows –Cognitive constructs correspond with emotions: anger, fear, and sadness Conduct statistical analysis on relationship between risk signals (IV) and affect pulse (DV) Analyze national longitudinal survey data over three time periods
12 Research Transition Plan: Complete analysis Summarize findings and recommendations Compile a brief report with recommendations for the use of Twitter data Possible briefing to TSA during summer 2015
13 Milestones and Schedule/Timeline: October 2014 – received funding at UK November 2014 – conducted literature review on risk perception, emotion/cognition, affect December 2014 – received funding at Decision Research December 2014 – conducted preliminary deductive analysis of 1000 randomly sampled tweets; identified risk signal pulses; identified anger/fear/sadness constructs; identified affect pulses January 2015 – select windows for qualitative coding; refine content analysis coding scheme February/March 2015 – conduct qualitative coding of content February/March 2015 – Top line summary of survey data April – conduct statistical analysis of both Twitter and Survey data. May/June 2015 – write up results/ present results