{"677744":{"#nid":"677744","#data":{"type":"news","title":"Study Shows Election Data Visualization Design Can Be a Powerful Persuasion Tool","body":[{"value":"\u003Cp\u003EFrom election forecasts and pandemic dashboards to stock market charts and scientific figures, many people trust data visualizations as objective truths and neutral representations of reality.\u003C\/p\u003E\u003Cp\u003EHowever, a study led by Georgia Tech and University of California, Berkeley researchers shows that annotations can lead people to draw different conclusions from the same visualizations. Their findings suggest readers should look beyond the presented data to make informed decisions.\u003C\/p\u003E\u003Cp\u003E\u201cPeople question things less if they see something that\u2019s visualized, and they think this is a reliable, trustworthy source they can use to form an opinion or persuade others,\u201d said Cindy Xiong, an assistant professor in the School of Interactive Computing. \u201cPeople don\u2019t realize the persuasive power of visualization, and they\u2019re not as vigilant to critically think about the data they interact with.\u201d\u003C\/p\u003E\u003Cp\u003EFor example, people tend to trust the information in an election data visualization. That makes them susceptible to narratives that visualization designers may use to obtain a certain outcome.\u003C\/p\u003E\u003Cp\u003EWorking with Chase Stokes, a Ph.D. candidate at UC Berkeley\u2019s School of Information, Xiong investigated how text position, semantic content, and biased wording impact viewers\u2019 perception of visualizations.\u003C\/p\u003E\u003Cp\u003EThey found people often reach the same conclusions suggested by titles and annotations on a chart.\u003C\/p\u003E\u003Cp\u003E\u201cVisual changes have a great deal of impact on how people interpret a chart,\u201d Stokes said. \u201cTitles, captions, and annotations strongly affect people\u2019s conclusions.\u201d\u003C\/p\u003E\u003Cp\u003EXiong and Stokes created a study centered around two hypothetical political parties \u2014 a blue party and a green party. They used a bar chart to show how many votes each party has received over the past three years. The data shows the blue party received more votes year after year than the green party, but the gap has closed each year.\u003C\/p\u003E\u003Cp\u003EThe researchers surveyed participants to predict which party would win in the fourth year. Responses were split nearly 50-50 before leveraging highlights and annotations.\u003C\/p\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cdiv\u003E\u003Cp\u003EWhen the researchers highlighted the green party\u2019s increasing voter support year after year, the prediction responses overwhelmingly favored the green party. Predictions favored the blue party when the researchers highlighted blue had won every year.\u003C\/p\u003E\u003Ch4\u003E\u003Cstrong\u003EExisting Bias\u003C\/strong\u003E\u003C\/h4\u003E\u003Cp\u003EWhile the scenario created by Xiong and Stokes reflects an ideal world of neutrality, the researchers concede that existing beliefs about political parties play a strong role in determining real-world bias. Participants consistently reported charts that supported one of the two parties were biased, and that perception intensified if the participants disagreed with the text provided.\u003C\/p\u003E\u003Cp\u003E\u201cIf I supported the green party, and I saw this chart, I would think blue party supporters made it because it\u2019s supporting the side that I don\u2019t agree with,\u201d Stokes said.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u201cIf the chart represented Republicans and Democrats, many people would perceive the data in a way that reinforces what they already think. If they disagreed with the party\u2019s ideologies, they would likely see the visualization as biased regardless of its portrayal.\u201d\u003C\/p\u003E\u003Ch4\u003E\u003Cstrong\u003EDesigner Responsibility\u003C\/strong\u003E\u003C\/h4\u003E\u003Cp\u003EXiong and Stokes found that although textually annotated data patterns do not strongly sway people\u2019s predictions to favor one party over another, they make people suspicious of the designer\u2019s beliefs.\u003C\/p\u003E\u003Cp\u003E\u201cIt\u2019s easy to make a chart that alienates half the people you\u2019re trying to reach,\u201d Stokes said. \u201cFiguring out a way to make data accessible, understandable, and interesting to people who may not agree with your story is critical to mending that trust between designer and consumer.\u201d\u003C\/p\u003E\u003Cp\u003EFor example, someone who trusts the information presented to them on Fox News may not trust what they see in The New York Times. Designers must account for the distrust between the public and information sources when creating their visualizations.\u003C\/p\u003E\u003Cp\u003E\u201cThe solution to reaching the widest possible audience is to provide both sides of the story, even if the designer wants to persuade people toward a certain perspective,\u201d Xiong said.\u003C\/p\u003E\u003Cp\u003E\u201cIf you are making visualizations for a political candidate, it\u2019s difficult to persuade people that you\u2019re not biased. You could visually highlight your key takeaways. But adding textual annotations to your chart will make people think you\u2019re pushing hard for some narrative.\u201d\u003C\/p\u003E\u003Ch4\u003E\u003Cstrong\u003EStaying Informed\u003C\/strong\u003E\u003C\/h4\u003E\u003Cp\u003EVoters, meanwhile, should be aware that most visualizations contain bias. The researchers agreed voters should gather information from various sources, including those that don\u2019t align with their opinions.\u003C\/p\u003E\u003Cp\u003E\u201cVoters should look for visualizations that talk about both sides and give you those different perspectives so you can make informed decisions about your future,\u201d Stokes said. \u201cIf you see a visualization that highlights one story, you should respond by finding the other side. There\u2019s never just one interpretation of a visualization.\u201d\u003C\/p\u003E\u003Cp\u003EXiong and Stokes published their findings in a paper that is being presented this week during the Institute of Electrical and Electronics Engineers\u2019 Visualization and Visual Analytics (VIS) Conference.\u003C\/p\u003E\u003Cdiv\u003E\u003Ch4\u003E\u003Cstrong\u003ERecent St\u003C\/strong\u003E\u003C\/h4\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E\u003C\/div\u003E","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003EFrom election forecasts and pandemic dashboards to stock market charts and scientific figures, many people trust data visualizations as objective truths and neutral representations of reality.\u003C\/p\u003E\u003Cp\u003EHowever, a study led by Georgia Tech and University of California, Berkeley researchers shows that annotations can lead people to draw different conclusions from the same visualizations. Their findings suggest readers should look beyond the presented data to make informed decisions.\u003C\/p\u003E","format":"limited_html"}],"field_summary_sentence":[{"value":"A study led by Georgia Tech and University of California, Berkeley researchers shows that annotations can lead people to draw different conclusions from the same visualizations."}],"uid":"36530","created_gmt":"2024-10-18 20:19:54","changed_gmt":"2024-10-18 20:20:50","author":"Nathan Deen","boilerplate_text":"","field_publication":"","field_article_url":"","dateline":{"date":"2024-10-18T00:00:00-04:00","iso_date":"2024-10-18T00:00:00-04:00","tz":"America\/New_York"},"extras":[],"hg_media":{"675369":{"id":"675369","type":"image","title":"2X6A2880.jpg","body":null,"created":"1729282801","gmt_created":"2024-10-18 20:20:01","changed":"1729282801","gmt_changed":"2024-10-18 20:20:01","alt":"Cindy Xiong","file":{"fid":"258982","name":"2X6A2880.jpg","image_path":"\/sites\/default\/files\/2024\/10\/18\/2X6A2880.jpg","image_full_path":"http:\/\/hg.gatech.edu\/\/sites\/default\/files\/2024\/10\/18\/2X6A2880.jpg","mime":"image\/jpeg","size":86109,"path_740":"http:\/\/hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/2024\/10\/18\/2X6A2880.jpg?itok=X6tNDuPV"}}},"media_ids":["675369"],"groups":[{"id":"47223","name":"College of Computing"},{"id":"1188","name":"Research Horizons"},{"id":"50876","name":"School of Interactive Computing"}],"categories":[{"id":"153","name":"Computer Science\/Information Technology and Security"}],"keywords":[{"id":"193818","name":"2024 Presidential election"},{"id":"193821","name":"2024 election"},{"id":"4065","name":"election"},{"id":"33301","name":"data analytics"},{"id":"38921","name":"data visualization"},{"id":"4508","name":"political"},{"id":"187915","name":"go-researchnews"},{"id":"9153","name":"Research Horizons"}],"core_research_areas":[{"id":"39501","name":"People and Technology"}],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[{"value":"\u003Cp\u003ENathan Deen\u003C\/p\u003E\u003Cp\u003ECommunications Officer\u003C\/p\u003E\u003Cp\u003ESchool of Interactive Computing\u003C\/p\u003E","format":"limited_html"}],"email":[],"slides":[],"orientation":[],"userdata":""}}}