Interactive visualization of electoral vote distributions comparing experimental and control groups across different news sources (MSNBC, FOX, BBC, and DIRECT). Each boxplot represents daily electoral vote totals for Kamala Harris, calculated by aggregating state-level win proportions weighted by electoral votes. Control groups (blue) had 46 days of data collection while experimental groups (green) had 81 days. The dashed line at 270 votes represents the threshold needed to win the Electoral College. Hover over any boxplot to see detailed statistics including the number of days, average trials per day, and vote distribution metrics. Outliers (top and bottom 1% per group) have been removed to better highlight the central tendencies in the data. This visualization is part of a larger study examining how different news source presentations affect electoral predictions.

Daily: 1. Pull 100 news stories from Event Registry API       • Prompt: “2024 US presidential election” (no date range) 2. Chat-GPT-4o-mini uses the news to create election-result stories in 4 distinct voices       • Four characters, each representing different perspectives, will generate 100 stories: • Voice 1: Anonymous/Direct reporter • Voice 2: Fox Reporter Bret Baier • Voice 3: MSNBC Reporter Rachel Maddow • Voice 4: BBC Reporter Laura Kuenssberg 3. Generate election stories from each character’s perspective       • For each character, 100 stories are written about the election outcome in each state. 4. Extract the election winners from each story       • Use GPT to extract only the name of the winners from the stories for each character. 5. Save winners in a matrix       • Store winners as: • 1 = Trump/Republican • 0 = Harris/Democrat           • Create a matrix with 100 trials for each of the 50 states. 6. Run the steps daily       • Repeat the process daily, appending new results to the previous day’s data panel. 7. Present the results in a Shiny app       • Display the results in the app, showing: • Daily winner for each state and overall. • Percentage of daily trials that went to each party in each state.

Section 2 still needs work. I haven't figured out how to add attributes/traits to clusters yet. I'm not sure I understand what he is asking for with size wither.