The purpose of this project is to compare media coverage of the National Basketball Association (NBA) and the Women’s National Basketball Association (WNBA) using aggregated story counts collected from Media Cloud. By examiining the volume of news coverage received by each league over time, the project provides an initial assessment of differences in media attention between men’s and women’s professional basketball.
Data were obtained from Media Cloud, an open-source platform that collects and analyzes online news content. Story counts were collected from April 19, 2025, through October 21, 2025.
The following search queries were used:
NBA
""LeBron James" OR "Shai Gilgeous-Alexander" OR NBA"
WNBA
""A'ja Wilson" OR "Caitlin Clark" OR WNBA"
Daily story counts were retrieved through the Media Cloud Application Programming Interface (API) and later aggregated into weekly totals.
After running the code, it produced two primary outputs:
The first output was an interactive Plotly line graph that illustrated weekly media coverage volume for the NBA and WNBA. The red line represented story counts for the NBA, and the blue line represented story counts for the WNBA.
The second output was a summary statistics table showing total stories, minimums, maximums, medians, means, and standard deviations for each league.
Together, these outputs allow users to examine overall coverage volume and how coverage changed throughout the study period.
I defined a problem, recognizing a need to compare weekly coverage volume between two professional sports leagues, specifically the NBA and the WNBA, that could be shared with other individuals who are interested in comparing coverage between two topics, and use this project as a base. Furthermore, this project highlights an ongoing problem the WNBA faces: even though it has been growing in audience participation and national viewership, it still doesn’t receive the same level of media coverage as the NBA.
I selected Media Cloud as the data source because it provides access to large collections of online news content.
I developed search queries, these seach queries were created to
represent both leagues. To identify NBA-related coverage, the search
query included terms such as LeBron James,
Shai Gilgeous-Alexander, and NBA. The search
term NBA was utilized as it is the name of the league.
LeBron James and Shai Gilgeous-Alexander were
selected because they are both prominent NBA players. Subsequently, the
WNBA query included A'ja Wilson,
Caitlin Clark, and WNBA, as the latter is the
league’s name. Additionally, A'ja Wilson and
Caitlin Clark were selected because they, too, are
prominent players, but within the WNBA.
I collected story-count data from the Media Cloud API
I aggregated the daily story counts into weekly totals.
After running the code, I created summary statistics and visualizations to display the results.
I reviewed ad refined the project outputs to improve clarity and presetation.
While artificial intelligence (AI) assisted with troubleshooting, code revisions, and documentation, decisions regarding the project topic, search terms, data source, analysis design, and interpretation remained my responsibility.
During this project, I gained experience working with APIs, automated date collection, and data visualizationin R Studio.
These are the specific skills I developed during the process of this project:
Accessing data through an API using the httr2
package.
Cleaning and transform data using tidyverse
functions.
Aggregating daily observations into weekly measures using
lubridate.
Creating interactive data visualizations with
Plotly.
Exporting datasets for additional analysis.
Documenting a reproducible workflow using R Markdown.
The project also demonstrated how computational tools can be used to collect and analyze media coverage at a scale that would be difficult to achieve through manual data collection.
The script begins by identifying the R packages needed for the
analysis, including tidyverse, httr2,
lubridate, plotly, knitr, and
kableExtra. It checks whether each package is installed and
installs any missing packages automatically. Once all required packages
are available, they are loaded into the R session.
Before collecting data, the script checks whether a valid Media Cloud
API key is stored in the user’s system environment. If no API key is
found, the script stops and displays an error message. If a key is
available, it is stored in the object mc_key for use in API
requests.
The script specifies the time period to be analyzed. For this project, data were collected between April 19, 2025, and October 21, 2025. These dates are later supplied to the Media Cloud API when requesting story counts.
The script creates a set of search queries that represent the topics being studied. For this project, the analysis focuses on the National Basketball Association (NBA) and the Women’s National Basketball Association (WNBA). The NBA query searches for stories containing “LeBron James,” “Shai Gilgeous-Alexander,” or “NBA,” while the WNBA query searches for stories containing “A’ja Wilson,” “Caitlin Clark,” or “WNBA.”
A custom function, mc_counts_over_time(), is created to
communicate with the Media Cloud API. This function submits a search
query along with the selected dates and collection information. Media
Cloud returns daily story-count data, which are converted into a data
frame for analysis.
The script loops through each search query and downloads daily story-count data from Media Cloud. Each set of results is labeled according to its topic (NBA or WNBA). A 35-second delay is included between requests to avoid exceeding Media Cloud’s API rate limits.
After the data are downloaded, the script displays the first several rows and the structure of the data set. This allows the user to verify that the download completed successfully and that the variables are formatted correctly.
The downloaded daily data are aggregated into weekly totals. First,
each date is assigned to a calendar week using the
floor_date() function. The script then groups observations
by topic and week and calculates the total number of stories that
occurred during each week.
The weekly data set is displayed so that the user can inspect the aggregated results before proceeding to summary statistics and visualization.
The script calculates descriptive statistics for each topic. These
statistics include the total number of stories, weekly minimum, weekly
maximum, weekly median, weekly mean, and weekly standard deviation. The
results are then formatted into a professional table using the
knitr and kableExtra packages.
An interactive line graph is created using the plotly
package. The graph displays weekly story counts for the NBA and WNBA
over time. Users can interact with the graph by zooming, panning, and
using the built-in range slider to focus on specific periods.
The weekly story-count data are exported as a CSV file named
weekly_story_counts.csv. This file can be opened in
spreadsheet software such as Microsoft Excel or imported into other
statistical programs.
The final step restructures the data into a wide format. Instead of
having separate rows for NBA and WNBA observations each week, the data
are reorganized so that each week appears once with separate columns for
NBA and WNBA story counts. This data set is then exported as
weekly_story_counts_wide.csv.
############################################################
# STEP 1: Install and Load Required Packages
# This step installs any missing packages and then
# loads the packages needed for the analysis.
############################################################
required_packages <- c(
"tidyverse",
"httr2",
"lubridate",
"plotly",
"knitr",
"kableExtra"
)
for (pkg in required_packages) {
if (!requireNamespace(pkg, quietly = TRUE)) {
install.packages(pkg)
}
}
library(tidyverse)
library(httr2)
library(lubridate)
library(plotly)
library(knitr)
library(kableExtra)
############################################################
# STEP 2: Verify Media Cloud API Key
############################################################
if (!nzchar(Sys.getenv("MEDIACLOUD_KEY"))) {
stop(
"No Media Cloud API key was found.\n",
"Run the setup section at the top of the script once,\n",
"then restart R and try again."
)
}
mc_key <- Sys.getenv("MEDIACLOUD_KEY")
############################################################
# STEP 3: Define the Date Range
# Enter the beginning and ending dates for the
# period you want to analyze.
############################################################
start_date <- "2025-04-19"
end_date <- "2025-10-21"
############################################################
# STEP 4: Define Your Topics
# Each topic consists of a topic label and a Media
# Cloud search query.
############################################################
# Agenda-setting query example:
topics <- c(
"NBA" =
"\"LeBron James\" OR \"Shai Gilgeous-Alexander\" OR NBA",
"WNBA" =
"\"A'ja Wilson\" OR \"Caitlin Clark\" OR WNBA"
)
# Framing query example:
# topics <- c(
# "Economic Frame" =
# "(immigration OR migrant) AND
# (jobs OR labor OR economy)",
#
# "Crime Frame" =
# "(immigration OR migrant) AND
# (crime OR drugs OR gang)",
#
# "Humanitarian Frame" =
# "(immigration OR migrant) AND
# (refugee OR asylum OR family)"
# )
############################################################
# STEP 5: Create a Function to Retrieve Daily
# Story Counts
# This function sends a query to Media Cloud and
# retrieves daily story counts.
############################################################
mc_counts_over_time <- function(
query,
start_date,
end_date,
collection_id = 34412234
) {
resp <-
request(
"https://search.mediacloud.org/api/search/count-over-time"
) |>
req_headers(
Authorization = paste("Token", mc_key)
) |>
req_url_query(
q = query,
start = start_date,
end = end_date,
platform = "onlinenews-mediacloud",
cs = collection_id
) |>
req_perform()
results <- resp |> resp_body_json()
map_dfr(
results$count_over_time$counts,
as_tibble
) |>
mutate(
date = as.Date(date)
)
}
############################################################
# STEP 6: Download Daily Counts for All Topics
# This step retrieves daily counts for every topic
# defined in the topics object.
#
# Media Cloud limits API requests, so a pause is
# inserted between queries.
############################################################
topic_data <-
imap_dfr(
topics,
function(search_string, topic_name) {
message("Downloading: ", topic_name)
results <-
mc_counts_over_time(
query = search_string,
start_date = start_date,
end_date = end_date
)
Sys.sleep(35)
results |>
mutate(
topic = topic_name
)
}
)
############################################################
# STEP 7: Examine the Downloaded Daily Data
# This step displays the first few rows of the
# downloaded data.
############################################################
head(topic_data)
glimpse(topic_data)
############################################################
# STEP 8: Convert Daily Counts to Weekly Counts
# This step aggregates daily story counts into
# weekly totals for easier visualization.
############################################################
weekly_counts <-
topic_data |>
mutate(
week = floor_date(
date,
unit = "week"
)
) |>
group_by(
topic,
week
) |>
summarise(
stories = sum(count),
.groups = "drop"
)
############################################################
# STEP 9: Examine Weekly Counts
# This step displays the weekly data that will be
# graphed.
############################################################
weekly_counts
############################################################
# STEP 10A: Summarize Coverage by Topic
# This step calculates descriptive statistics for
# weekly story counts and displays the results in
# a formatted table.
############################################################
topic_summary <-
weekly_counts |>
group_by(topic) |>
summarise(
Total_Stories = sum(stories),
Weekly_Minimum = min(stories),
Weekly_Maximum = max(stories),
Weekly_Median = median(stories),
Weekly_Mean = round(mean(stories), 2),
Weekly_SD = round(sd(stories), 2),
.groups = "drop"
) |>
arrange(desc(Total_Stories))
Table <-
topic_summary |>
kbl(
caption = "Media Cloud Topic Summary Statistics",
align = "lrrrrrr"
) |>
kable_styling(
full_width = FALSE,
bootstrap_options = c(
"striped",
"hover",
"condensed"
)
)
Table
############################################################
# STEP 10B: Create an Interactive Plotly Graph
# This step creates a line graph showing weekly
# story volume for each topic.
############################################################
Graph <- plot_ly(
data = weekly_counts,
x = ~week,
y = ~stories,
color = ~topic,
colors = "Set1",
type = "scatter",
mode = "lines+markers"
) |>
layout(
title = "Weekly U.S. Coverage Volume",
xaxis = list(
title = "Week (with range slider)",
rangeslider = list(visible = TRUE)
),
yaxis = list(
title = "Weekly Story Count"
)
)
Graph
############################################################
# STEP 11: Save Weekly Counts as a CSV File
# This step exports the weekly counts to a CSV file
# that can be opened in Excel or imported into
# other software.
############################################################
write_csv(
weekly_counts,
"weekly_story_counts.csv"
)
############################################################
# STEP 12: Create a Wide-Format Version
# This step creates a spreadsheet with one row per
# week and one column for each topic.
############################################################
weekly_counts_wide <-
weekly_counts |>
pivot_wider(
names_from = topic,
values_from = stories
) |>
arrange(week)
write_csv(
weekly_counts_wide,
"weekly_story_counts_wide.csv"
)
weekly_counts_wide