The National Endowment for the Humanities (NEH) provides an extensive online repository of historical data. Notably, this data set makes it possible to study the allocation of funding over every decade of the endowment’s existence. Taking three distinct angles—chronological, spatial, and categorical—this document presents findings on the distribution of funding to answer the question, “Where does NEH funding go?”
# Load packages used within this document
library(tidyverse)
library(scales)
library(quarto)
# Load functions and data prepared in other scripts
source("R/parse_data.R")
source("R/mapping.R")Available in XML files, the NEH funding data for each decade reveals which projects were awarded funding, how much they were awarded, and additional details such as their project title, field of study, and location. To study spending over time, some values need to be added for context. First, comparing monetary values over many years only makes sense if the value of a dollar is stable; for this reason, real dollar values have been indexed to 2022 dollars using the priceR package in R, which works from World Bank inflation data. Second, larger states will reasonably show higher funding values than smaller states; accordingly, historical census estimates for each state in each year have been imported to calculate per capita award values. Adopting each method substantially changes what the data reveal.
In the charts that follow, award amounts are calculated as the combination of each project’s AwardOutright and AwardMatching values in the NEH-supplied data and then combined by state and year. Colors are determined by a state’s median value on the Y-axis. Color neighbors suggest that two states share similar funding history, while a lack of nearby colors shows that a state or district may be an outlier.
all_decades3 |>
plot_overview(
award_total,
"Nominal spending per state") +
# Change the color scheme
scale_color_distiller(
direction = 1,
palette = "YlGn") +
# Add a trend line
geom_smooth(se = FALSE)
Reported numbers suggest an exclusive club, typically reserving annual funding amounts over $10 million to states like New York, California, and Massachusetts. Other states tangle at the bottom with a trend line around $2.5 million per state staying relatively stable after 1980.
all_decades3 |>
plot_overview(award_adjusted, "Real spending per state (2022 dollars)") +
# Use the same color scheme for dollars per state
scale_color_distiller(
direction = 1,
palette = "YlGn") +
# Add a trend line
geom_smooth(se = FALSE)
Adjusting for inflation shows a different shape in the data. Here, funding rises to a sharp peak in the late 1970s, at a time when New York secures over $100 million in grant funding for four out of five years. Nationally, the trend line shows a continual decline since that point except for a blip of renewed funding around 1990.
all_decades3 |>
plot_overview(per_capita_adjusted, "Real spending per capita (2022 dollars)") +
# Use a different color scheme for different unit (dollars per capita)
scale_color_distiller(
direction = 1,
palette = "YlGnBu") +
# Add a trend line
geom_smooth(se = FALSE)
Adjusting annually for the estimated population of each state or district adds further context. The District of Columbia dominates per capita funding, with amounts commonly double that of the next-highest state.
all_decades3 |>
filter(state != "DC") |>
plot_overview(per_capita_adjusted, "Real spending per capita (without DC)") +
# use the dollars per capita color scheme
scale_color_distiller(
direction = 1,
palette = "YlGnBu") +
# Add a trend line
geom_smooth(se = FALSE)
Omitting DC rescales the chart to reveal other states with higher amounts of per capita funding. And similar colorations show many states to have comparable median values. States with higher values of per capita funding tend to have smaller populations, including Alaska, Wyoming, Delaware, and Vermont.
Bigger states have gotten bigger amounts of each year’s funding, but smaller states earned larger portions per capita. Standardized for inflation, annual funding amounts have fallen steadily since about 1990.
Projecting data to a map makes it possible to find any regional trends. To make these maps, it is necessary to have workable shape files describing the contours of each state. The tigris package in R supplied this state shape data. After geospatial data was imported for each state, Alaska and Hawaii were transformed to appear under the continental 48 states to simplify comparison.
awards_map_final |>
map_years(1966:1971)
The first few years of data show a slow start, an undeniable interest in national coverage, and a few rocky years. In 1966, one year after its founding, the NEH funded 10 projects on the east coast, limited to New York, Connecticut, and Washington DC. It funded 612 projects in the next year, with representation stretching across the country to include DC and every state but New Mexico. In 1968, funding dropped dramatically, with just 149 awards, but it recovered in a few years.
Comparing these maps suggests an early interest in gaining geographic diversity. Per capita funding was comparatively low in southern states—and indeed in most states away from the east and west coasts—but this regional pattern does not seem to last long.
awards_map_final |>
map_year(1977)
Funding peaked in 1977, when more than two thousand projects were awarded $721.5 million, adjusted for inflation. DC’s projects were awarded more than $47 per capita, and Alaska’s topped $15 per capita (2022 dollars). As this map shows, no region dominates funding dollars, and none seems to suffer from poor representation.
awards_map_final |>
map_year(2019)
In 2019, funding fell to its lowest level since the 1960s. Only 424 projects were funded, totaling $60.6 million when adjusted for inflation. Per capita funding averaged just 33¢ in states where projects were funded. While every state suffers in lean times like this, they seem to do so unequally. New England states, in particular, seem to have weathered the funding crunch better than most. Dark zones along the Ohio River, the Mississippi River, and the border with Mexico show regions of states with noticeably lower funding. White blanks show some states that lack funding altogether.
Maps show good geographic diversity, especially in good years of funding. In bad years, certain regions may feel it harder than others. The Northeast fares better than other regions when funding is low.
Finally, the NEH data sets provide sufficient opportunity for categorical comparisons over time and against different groups.
all_decades <- readRDS("data/all_decades.rds")
first_timers <- all_decades |>
arrange(year) |>
filter(
!str_detect(institution, "Unknown"),
!str_detect(institution, "Unaffiliated")) |>
group_by(institution) |>
mutate(num_awards = row_number(),
year = as.integer(year)) |>
ungroup() |>
summarize(
percent = sum(num_awards == 1) / n(),
.by = year)
first_timers |>
ggplot(aes(year, percent)) +
geom_point() +
geom_line() +
scale_x_continuous() +
labs(x = NULL,
y = NULL,
title = "Proportion of first-time awardees spiked in 1987 and 1999") +
scale_y_continuous(labels = label_percent()) +
theme_minimal() +
theme(plot.title.position = "plot")
Not all grants go to perennial winners of funding. As this chart shows, about 13% of all awards since 2000 have gone to first-time awardees. An interesting spike in 1987 and a rise in 1999-2000 are worth further study.
all_decades |>
filter(state %in% c(state.abb, "DC")) |>
summarize(total = sum(award_total, na.rm = TRUE),
.by = division) |>
ggplot(aes(total, reorder(division, total))) +
geom_col(fill = "black") +
scale_x_continuous(labels = label_dollar(scale_cut = cut_short_scale())) +
theme_minimal() +
theme(panel.grid.major.y = element_blank()) +
labs(x = NULL,
y = NULL,
title = "Divisions by total NEH award funding") +
theme(plot.title.position = "plot")
Although academics might think primarily of the research programs funded by the NEH, the largest portion of funding goes to federal / state partnerships. Education programs and public programs are also funded at higher rates than research.
all_decades |>
filter(state %in% c(state.abb, "DC")) |>
summarize(total = sum(award_total, na.rm = TRUE),
.by = discipline) |>
arrange(desc(total)) |>
head(10) |>
ggplot(aes(total, reorder(discipline, total))) +
geom_col(fill = "black") +
scale_x_continuous(labels = label_dollar(scale_cut = cut_short_scale())) +
theme_minimal() +
theme(panel.grid.major.y = element_blank()) +
labs(x = NULL,
y = NULL,
title = "Top-10 disciplines by total NEH award funding") +
theme(plot.title.position = "plot")
Compared to other disciplines, interdisciplinary studies has received far more funding over time. But this conclusion might be a matter of categorical convenience, since many projects involve partnerships of academics in multiple disciplines. History makes the chart three times, as does literature.
all_decades |>
filter(state %in% c(state.abb, "DC")) |>
summarize(total = sum(award_total, na.rm = TRUE),
.by = org_type) |>
arrange(desc(total)) |>
head(10) |>
ggplot(aes(total, reorder(org_type, total))) +
geom_col(fill = "black") +
scale_x_continuous(labels = label_dollar(scale_cut = cut_short_scale())) +
theme_minimal() +
theme(panel.grid.major.y = element_blank()) +
labs(x = NULL,
y = NULL,
title = "Top-10 organizational types by total NEH award funding") +
theme(plot.title.position = "plot")
Finally, the lion’s share of funding goes to state humanities councils, followed by four-year colleges and universities. From there, a long tail of organizational types show funding going to centers and institutes, national organizations and societies, and museums, among other places.
Comparing groups shows that funding goes to a broad mix. First-time recipients make up an impressive share of awardees, and four-year colleges are awarded more than universities. Research programs represent a smaller share than some other kinds of funding, and interdisciplinary work is the best funded.
Considering history, geography, and types of projects funded can show insights into the NEH’s funding practices. Studying many perspectives can help ensure that the endowment touches more lives, needlessly missing none.
This work was completed using R v. 4.5.0 (R Core Team 2025) and the following R packages: grateful v. 0.3.0 (Rodriguez-Sanchez and Jackson 2025), janitor v. 2.2.1 (Firke 2024), priceR v. 1.0.3 (Condylios 2025), quarto v. 1.4.4 (Allaire and Dervieux 2024), rmarkdown v. 2.29 (Xie, Allaire, and Grolemund 2018; Xie, Dervieux, and Riederer 2020; Allaire et al. 2024), scales v. 1.4.0 (Wickham, Pedersen, and Seidel 2025), sf v. 1.0.20 (Pebesma 2018; Pebesma and Bivand 2023), tidyverse v. 2.0.0 (Wickham et al. 2019), tigris v. 2.2.1 (Walker 2025). Code for the report can be viewed alongside its supporting materials on GitHub at the following URL: https://github.com/jmclawson/neh_data/blob/postsubmit/report.qmd