Social Data Collaboratory

At the Social Data Collaboratory, we've been exploring Gen-AI for text classification. We tasked GPT-4 with classifying two data sources: social media posts on abortion (GSMA) into Pro-Choice, Pro-Life, and Neutral attitudes; and AP-NORC open-ended survey responses on key political topics (gun issues, abortion, immigration, environment, economy). We compared GPT-4's accuracy against classical machine learning models trained on Azure AutoML. Key Takeaways: ✅ GPT-4 outperforms AutoML models. ✅ Near 100% accuracy for AP-NORC open-ended responses. ✅ Nearly 20% improvement on GSMA abortion posts. Our findings highlight GPT-4's potential in text classification tasks. 🌟 NORC at the University of Chicago #GenAI #MachineLearning #TextClassification #PublicHealth #Innovation 📊📈✨

New Research from NORC at the University of Chicago Social Data Collaboratory (SDC)! The research team includes Anna (Ganna) Kostygina, Sherry Emery, Simon Page. We're thrilled to share our latest findings on the impact of social media marketing on tobacco sales. We took a novel approach: using geolocation-specific tweet rates as an exogenous indicator of exposure to smokeless tobacco (ST) content and employed this measure to examine the influence of social media marketing on ST sales. Key Takeaways: Social Media Influence: A 100-unit increase in ST-related tweets was linked to a 14% rise in sales for newer products and a 1% rise for conventional products. Price Impact: Higher prices were negatively associated with ST sales, highlighting the interplay between cost and social media influence. Policy Control: Our model accounted for policy changes and the COVID-19 pandemic, ensuring robust results. These findings underscore the critical role social media plays in promoting tobacco use and the necessity of monitoring these platforms for emerging marketing strategies. By using exogenous measures like tweet rates, we can better understand and combat the influence of social media on public health. 🔗 Please check out full article at: https://lnkd.in/e-FBquhR 🤝 Call to Action: We're eager to collaborate with fellow researchers, public health professionals, and policymakers to further explore this dynamic landscape. Feel free to contact us to exchange ideas. Let's connect and make a difference together! 🌍📊 #PublicHealth #SocialMediaResearch #TobaccoControl #InnovativeResearch

New SDC research with Truth Initiative on youth and young adult awareness of perceptions of tobacco marketing as a social justice issue! NORC at the University of Chicago #PublicHealth #Tobacco #SocialJustice

🚀 Excited to share our recent work showcased at #WAPOR24 by Miao Feng, Haoyu Shi, and Simon Page in Seoul, South Korea last week! 1️⃣ Flavored Combustible Tobacco Product Initiation in Two Longitudinal Cohorts of Youth in the Population Assessment of Tobacco and Health Study: Our study reveals that social media significantly increases youth exposure to flavored tobacco promotions. E-cigarette use also lowers barriers to nicotine addiction, making flavors less crucial for initiating tobacco use. This underscores the need for more research on social media's role in promoting tobacco among youth. 2️⃣ Leveraging Social Media Data to Examine Public Abortion Discourse - A Case Study Using the General Social Media Archive: We found platform-specific trends: Facebook leans pro-life, Twitter leans pro-choice, and Instagram is mixed. Pro-life content shows greater volume and engagement, highlighting the organized nature of pro-life advocacy online. This research emphasizes the importance of social media in shaping public opinion on contentious issues like abortion. 3️⃣ Analyzing U.S. E-cigarette Prevention Campaigns on TikTok (2020-2023): Despite the limited presence of e-cigarette prevention campaigns on TikTok, their high engagement levels indicate a significant potential for impact. Amplifying these efforts could profoundly influence youth behavior on this popular platform. 📊 Our findings highlight the crucial role of social media in public health and opinion dynamics. NORC at the University of Chicago #PublicHealth #YouthSmoking #TobaccoPrevention #SocialMediaResearch #AbortionDebate #TikTok #HealthCommunication 🌍📈

📊 Wondering about the value of analyzing newer and niche platforms like Twitch, Discord, Mastodon, Gab, Gab, Gettr, Truth Social, BitChute, Rumble and others? NORC at the University of Chicago's Social Data Collaboratory (SDC) harnesses these platforms to uncover unique insights into evolving trends and behaviors. By exploring these emerging digital spaces, we provide a comprehensive understanding of public opinion and behavior that traditional methods might miss. Contact us to leverage these cutting-edge capabilities for strategic decision-making: https://lnkd.in/g55RzDzf #SocialData #DataAnalytics #ResearchInsights

More about social media! I thought this comment, in reaction to the proposed warning labels, was super interesting: "Dr. Odgers said the nation’s top health official was running the risk of labeling normal adolescent behavior as “shameful, damaging and dangerous.” Missing in this conversation is the pervasiveness and effects of undisclosed influencer marketing of commercial determinants of health--e.g., tobacco, alcohol, cannabis, prescription drugs, online gambling.

📢New research alert! At NORC at the University of Chicago's Social Data Collaboratory (SDC), we examined the efficacy of influential account metrics for social media studies, shedding light on crucial tools needed by public health and public opinion researchers to identify influential accounts. Key Takeaways: 📊Effective Metrics: We identified distinct metrics for measuring influential accounts on social media, crucial for understanding discussions around topics. 🌐Metric Diversity: Using a variety of metrics is essential to accurately capture different types of influential accounts and their impact on public health conversations. 🕰️Tailored Research Approaches: Researchers should consider the temporal dimension and account types captured by different metrics to design more effective studies in public health and social media. These insights highlight the importance of leveraging diverse metrics to gain comprehensive insights into how social media influencers shape public health discussions. 📑Read the full study here: https://lnkd.in/dE_Qb4ff 🤝Interested in collaborating on research to address public health challenges in the digital age? Contact us to explore innovative solutions and partnerships! #SocialData #DigitalInsights #ResearchAnalytics #SocialTrends

We're beginning to see a trade-off between creating paywalls for training data access and letting the AI companies just extract value from others' content. I'm all for honoring copyrights and having the AI developers pay for the valuable resource. However, one of the most concerning (to me) consequences of protecting data provenance is that academics and not-for-profit researchers face yet another paywall for access to data. There's already a divide between those who can afford access to data and those who cannot. It's only going to get worse unless the US adopts a new approach to digital regulation that protects public and research use applications.

Short-video social media platforms like #TikTok and #Reels are experiencing a surge in popularity. At NORC at the University of Chicago's Social Data Collaboratory (SDC), we are proactively embracing video research methodologies to stay ahead. Haoyu Shi has recently leveled up our capabilities by attending Video Research Institute at Purdue University, expanding our expertise from traditional text analyses to cutting-edge audio and video technologies. What's New? We're now equipped with advanced tools for: 🗣️ Vocal Pitch Analysis for Emotion 🔊 Wav2Vec Audio Embeddings 🫰 Gesture & Facial Expression Detection 🐇 Obejct Detection & Motion Tracking 🌐 Multimodal Deep Learning Join us in exploring the limitless possibilities of video research. Your insights could shape the future of how we interact with media! 🔗 Stay Connected, Stay Informed. #VideoResearch #Innovation #AI #MachineLearning #ResearchExcellence #Multimodal #DeepLearning #SocialMedia

It's so much easier to propagate disinformation than to correct or neutralize it because 'truth' is rarely categorical and is often interpreted through cultural context. It's fascinating that the misinformation machines can so easily identify those appeals, the two truths, that contextualize and make plausible the one lie. The challenge for researchers is to a) recognize the complexity, and b) to honor, rather than dismiss, the context that makes the disinformation salient. Only when we can do that, and do it at scale and with speed, will we be able to achieve progress. No small challenge!