This project analyzes how political leaders across six countries—Australia, Brazil, Canada, India, the United Kingdom, and the United States—used public communication to shape national sentiment during the COVID-19 pandemic from 2020 to 2022. Using a custom-built sentiment dataset of over 200 speeches, I created an interactive Crosstalk dashboard that visualizes patterns in tone, rhetorical volatility, descriptive language, and temporal shifts in messaging. Monthly and yearly sentiment trends reveal how national communication strategies diverged as countries navigated outbreaks, restrictions, vaccination campaigns, and periods of political uncertainty. Interactive filters allow users to isolate specific countries, themes, or time windows to explore how tone evolved in response to major events. In addition to the comparative analysis, the project includes a dedicated U.S. case study that reconstructs the narrative of American pandemic communication through detailed daily sentiment timelines, theme-level summaries, and keyword extraction. This provides a deeper look into how U.S. messaging transitioned from crisis response to vaccine rollout, variant surges, economic stabilization, and public-health fatigue. Overall, the purpose of this dashboard is to provide a research-ready foundation for understanding the relationship between political communication and public sentiment shifts during one of the most consequential global crises of the 21st century. The visualizations support a 2,500-word analytical paper by highlighting clear empirical patterns, contrasts between countries, and the rhetorical strategies used by governments as the pandemic evolved.

--- title: "U.S. COVID-19 Sentiment Timeline" output: html_document --- ```{r setup, include=FALSE} knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE) library(dplyr) library(ggplot2) library(lubridate) library(crosstalk) library(plotly) library(DT) set.seed(123) ``` ```{r load_and_prepare} # Load data ------------------------------------------------------------- # If the CSV is on your Desktop: sent <- read.csv("~/Desktop/usa_covid_2020_2022.csv", stringsAsFactors = FALSE) # If it's in the project folder instead, use: # sent <- read.csv("usa_covid_2020_2022.csv", stringsAsFactors = FALSE) sent$Date <- as.Date(sent$Date) sent <- sent[order(sent$Date), ] sent <- sent %>% mutate( Year = year(Date), Month = month(Date), YearMonth = format(Date, "%Y-%m") ) # Theme tagging --------------------------------------------------------- assign_theme <- function(text) { txt <- tolower(text) if (grepl("vaccine|vaccination|shot|booster|dose|warp speed", txt)) { "Vaccines & Boosters" } else if (grepl("lockdown|stay-at-home|shutdown|restrictions|social distancing|face covering|mask", txt)) { "Restrictions & Reopening" } else if (grepl("delta|omicron|variant|ba\\.5|ba5", txt)) { "Variants" } else if (grepl("economy|economic|jobs|unemployment|market|business|stimulus|checks|rescue plan|relief", txt)) { "Economy & Relief" } else if (grepl("school|schools|students|classroom|kids", txt)) { "Schools & Education" } else if (grepl("misinformation|disinformation|false information|myths|rumors", txt)) { "Misinformation & Communication" } else { "General Pandemic Response" } } sent$Theme <- vapply(sent$Summary, assign_theme, character(1)) # Stopword lists for word scrambles ------------------------------------- stopwords_simple <- c( "the","and","to","of","in","a","for","on","as","is","are","this","that", "with","be","from","at","it","by","an","into","about","has","have","will", "was","were","but","more","many","most","less","very","their","his","her", "its","our","us","we","they","over","again","now","than","still","even" ) stopwords_domain <- c( "covid","coronavirus","pandemic","virus","viruses", "case","cases","death","deaths", "vaccine","vaccines","shot","shots", "lockdown","lockdowns","restrictions","guidelines", "americans","american","people","country", "united","states","state","u","s","us","u.s", "economy","economic","hospital","hospitals" ) make_scramble <- function(text) { clean <- gsub("[^A-Za-z ]", " ", tolower(text)) words <- unlist(strsplit(clean, "\\s+")) words <- words[ words != "" & !(words %in% c(stopwords_simple, stopwords_domain)) & nchar(words) >= 5 ] if (length(words) == 0) return("") unique_words <- unique(words) paste(sample(unique_words, min(5, length(unique_words))), collapse = " ") } # Monthly aggregates ----------------------------------------------------- monthly_info <- sent %>% group_by(YearMonth) %>% summarize( MonthStart = min(Date), AvgSentiment = mean(SentimentScore), HoverText = paste(Summary, collapse = "<br>• "), MonthScramble = make_scramble(paste(Summary, collapse = " ")), .groups = "drop" ) %>% mutate( SentSign = ifelse(AvgSentiment > 0, "Positive", "Negative/Neutral") ) # Join month scramble back to row level --------------------------------- sent <- sent %>% left_join(monthly_info %>% select(YearMonth, MonthScramble), by = "YearMonth") # Theme-level summary --------------------------------------------------- theme_summary <- sent %>% group_by(Theme) %>% summarize( AvgSentiment = mean(SentimentScore), Count = n(), .groups = "drop" ) %>% mutate( SentSign = ifelse(AvgSentiment > 0, "Positive", "Negative/Neutral") ) ``` ```{r monthly_interactive} plot_ly( monthly_info, x = ~MonthStart, y = ~AvgSentiment, type = "scatter", mode = "lines+markers", color = ~SentSign, colors = c("Negative/Neutral" = "#d95f02", "Positive" = "#1b9e77"), text = ~paste0( "<b>", YearMonth, "</b><br>", "Average Sentiment: ", round(AvgSentiment, 3), "<br>", "Category: ", SentSign, "<br><br>", "<b>Summaries:</b><br>• ", HoverText ), hoverinfo = "text" ) %>% layout( title = "Monthly Average Sentiment (Color-Coded by Positive vs Negative)", xaxis = list(title = "Month"), yaxis = list(title = "Average Sentiment Score (-1 to +1)") ) ``` ```{r theme_chart} ggplot( theme_summary, aes(x = reorder(Theme, AvgSentiment), y = AvgSentiment, fill = SentSign) ) + geom_col(width = 0.6) + geom_hline(yintercept = 0, linetype = "dashed") + coord_flip() + scale_fill_manual( values = c("Negative/Neutral" = "#d95f02", "Positive" = "#1b9e77"), name = "Average Sentiment" ) + theme_minimal(base_size = 13) + labs( title = "Average Sentiment by Theme", x = "Theme", y = "Average Sentiment Score (-1 to +1)" ) ``` ```{r interactive_timeline} shared_sent <- SharedData$new(sent, key = ~Date, group = "covid") filter_date <- filter_slider( id = "date_filter", label = "Date Range", sharedData = shared_sent, column = ~Date ) filter_score <- filter_slider( id = "score_filter", label = "Sentiment Score", sharedData = shared_sent, column = ~SentimentScore, step = 0.05 ) filter_month <- filter_select( id = "month_filter", label = "Month", sharedData = shared_sent, group = ~YearMonth ) timeline_plot <- plot_ly( shared_sent, x = ~Date, y = ~SentimentScore, type = "scatter", mode = "lines+markers", text = ~paste0( "Date: ", Date, "<br>", "Sentiment: ", SentimentScore, "<br>", "Theme: ", Theme, "<br><br>", Summary ), hoverinfo = "text" ) %>% layout( title = "Daily Sentiment Timeline", xaxis = list(title = "Date"), yaxis = list(title = "Sentiment Score (-1 to +1)") ) bscols( widths = c(3, 9), list(filter_date, filter_score, filter_month), timeline_plot ) ``` ```{r word_scramble_table} table_data <- sent %>% select(Date, SentimentScore, YearMonth, MonthScramble, Theme, Summary) shared_table <- SharedData$new(table_data, key = ~Date, group = "covid") datatable( shared_table, options = list(pageLength = 5, lengthChange = FALSE), colnames = c( "Date", "Sentiment", "Month", "Word Scramble", "Theme", "Summary" ), rownames = FALSE ) ```