Although the present research was exploratory, there are good theoretical reasons to expect relationships between Section 117 contributions and illiberal trends in American institutions of higher education. We examined two potential aspects of campus life in which receipt of Section 117 contributions may be involved: 1. erosion of free speech and academic freedom, and 2. antisemitism.

How might this work within universities? First, Section 117 money may be used to facilitate an intolerant intellectual environment on campus, one with higher levels of deplatforming, more calls for punishing speech and expression, and endorsement of censorship. Second, the money might be used to support and expand the work of faculty or students who are willing to violate others' speech rights. Third, the money might be used to support antisemitic extremist groups on campus and/or groups whose activities spill over into violation of others' speech rights. Fourth, it might buy a sort of soft influence—the institutional administration or whoever is receiving the funds might become more sympathetic to the interests and ideological agendas of the donor.

The above analysis is speculative and describes the types of processes that should be further investigated, pending empirical answers to the research questions driving the present set of studies. Nonetheless, there are some real-world reasons to believe such processes may already be occurring. For example, the Communist Party of China (CCP) funds Confucius Institutes throughout higher education, ostensibly to advance the study of Chinese languages, history and culture. In fact, however, these institutes constitute a major CCP propaganda effort, which includes suppressing speech and discourse around issues embarrassing to the CCP (e.g., Uyghurs, Tiananmen Square, Tibet and Taiwan; Human Rights Foundation, 2021). The research reported herein can be viewed as a preliminary attempt to determine whether this type of report constitutes anything more than a rare anecdote or something more pervasive and systematic.

The flood of Section 117 funding described herein coincides with both increased illiberal, anti-democratic sentiment on American college and university campuses (Rauch, 1993; Stevens, 2022, 2023) and antisemitic incidents (Beckwith and Rossman-Benjamin, 2022). The research reported herein provides the first effort to quantitatively examine the potential relationship between Section 117 funds, anti-democratic trends, and levels of antisemitism on American college and university campuses. Specifically, the present work investigates the following questions:

The data sources drawn upon include:

We employ simple, straightforward definitions of antisemitism and anti-Zionism in the present paper. Antisemitism refers to prejudice or discrimination against Jews, individually or collectively (for related definitions and perspectives, see Allington et al., 2023; Cohen et al., 2009; Kaufman et al., 2020). This may manifest in a myriad of ways, including but not restricted to open hostility, covert hostility, belief in conspiracy theories involving Jews, holding unjustified derogatory beliefs about Jews, hate crimes against Jews, harassing people because they are Jewish, etc.

Zionism refers to the movement to establish and maintain a Jewish state in Israel. As such, anti-Zionism refers to hostility toward and opposition to Israel as a Jewish state. Antisemitism and anti-Zionism, therefore, are, conceptually, clearly distinguishable. Empirically, however, they have been found to correlate at about r = 0.40 (Allington et al., 2023; Cohen et al., 2009) which means there is about a 70% chance that, people who score in the upper half of one measure, will also score in the upper half of the other (Rosenthal and Rubin, 1982).

For incidents of antisemitism on campus, in this report, we use the term “Expression” to refer to AMCHA's recording of expressions of antisemitism that do not target particular students or Jewish institutions, such as episodes of antisemitic graffiti, slogans, and chants. We use the term “Targeting” to refer to AMCHA's reporting of incidents of antisemitism on campus that target specific students (e.g., harassment) and institutions (e.g., defacing a Hillel). Some analyses also rely on AMCHA data on activities involving the anti-Zionist Boycott, Divestment, and Sanctions (BDS) movement on campus. The question of whether BDS activities are antisemitic is contentious. Resolving this question is beyond the scope of the present paper. Therefore, we present anything involving BDS as “antisemitic/anti-Zionist” to capture the idea that it is one, the other, or both. The empirical question relevant to the present research is whether foreign funding is associated with heightened BDS activity, and we leave it to others to decide for themselves whether such activity is antisemitic, anti-Zionist, or both. Determining correlations between support for BDS and hostility toward Jews was beyond the scope of the present studies, though it is an interesting and important question for future research.

From 1981 to 2020, the DoEd hosted a portal for foreign funding of American institutions of higher education. This portal made publicly available the amount of foreign funding received by those institutions as they reported it to the DoEd. From 2014 to 2019, Michael Bass downloaded the information as it became available (twice per year) and consolidated the new reporting information with all previous information. Starting in 2020, the DoEd changed the portal mechanism, so that it now includes all information since 1981 (U.S. Department of Education, n.d.).

Between 2014 and 2019, 349 U.S. colleges and universities received Section 117 funding from 158 foreign countries. The total sum across all universities and years was $17,961,986,172. Table 1A shows the top 10 countries contributing this type of funding to U.S. institutions of higher education during this time period. Table 1B shows the 10 universities that received the most Section 117 funding during this time period. Supplementary Tables provide full data on all countries and institutions. Supplementary Table S1 reports how much Section 117 funding had been donated by each of the 158 countries to U.S. institutions of higher education. Supplementary Table S2 lists the amounts of Section 117 funding received by each of the 349 institutions of higher education during this time period. The values obtained in Study I (see Supplementary Table S2) are used in all subsequent studies assessing the relationship of Section 117 funding to other outcomes.

This study assessed relations between receipt of Section 117 funds and FIRE data on Scholars Under Fire and campus deplatforming attempts (FIRE, n.d.-a,b) for 2014–2023.¹

The relevant information from FIRE databases that we included was as follows:

Supplementary Table S3 provides the list of OIC member countries that provided Section 117 funding. Supplementary Table S4 provides the list of authoritarian countries that provided Section 117 funding.

There were 986 Scholars Under Fire and 771 deplatforming attempts between 2014 and 2023 (summarized by year in Table 2).

Table 6 summarizes results of Welch's t-tests comparing experiences of campus antisemitism reported by students attending universities that received Section 117 funds between 2014 and 2019 (n = 473) with such experiences reported by students from universities that did not receive Section 117 funds (n = 1,287). First, the overall mean levels of reported exposure to antisemitic and anti-Zionist tropes were low; all means were below 1.4 on a scale of 1–4, where 1 was “never.” Consistent with this, the effect sizes are uniformly small. Second, even though overall reported exposure was low, students at colleges/universities that received Section 117 funding reported experiencing all five types of antisemitism more often, in terms of mean differences, than did students at institutions that did not receive Section 117 funding. However, only two had p-values below 0.01 (Israel has no right to exist; boycott), and a third (American Jews care more about Israel) had a p-value of 0.013. The remaining items did not approach statistical significance (see Table 6).

Additional analyses assessed whether experiences with antisemitism differed depending on whether funding was from OIC vs. non-OIC countries, and whether funding was from authoritarian vs. non-authoritarian countries. However, all of these differences were trivial in magnitude, and none had p-values below 0.05 (all Welch's t's < 1.0, all p's ≥ 0.30). Overall, therefore, Study III provided modest evidence of students reporting more exposure to antisemitic and anti-Zionist tropes if they attended an institution that received Section 117 funding than if they attended one that did not.

A limitation of Study III was that it assessed students' self-reported exposure to antisemitic and anti-Zionist rhetoric. Exposure to rhetoric and being a victim of an actual hate crime or some form of harassment or vandalism are very different types of phenomena. Study V examined these issues. However, before studying whether receipt of Section 117 funding was related to actual antisemitic incidents, Study IV assessed whether different national measures of antisemitic incidents assessed similar or fundamentally different underlying phenomena.

We obtained data on antisemitic acts and incidents from the websites of the FBI, ADL, and AMCHA for 2015–2020. Our first objective was to assess the consistency among their reports. Data on antisemitic activity in 3,108 U.S. counties in the lower 48 states was obtained from the FBI's Uniform Crime Reporting database for the years 2015–2020. We linked the AMCHA data on campus antisemitism to county by using Google map data for campuses. The ADL Tracker™ dataset contains data on antisemitic incidents for all 50 states. The ADL, AMCHA and FBI spreadsheets were converted from city form to county FIPS code using INDEX and MATCH functions in Excel with an extensive dataset of 108,797 cities and towns downloaded from SimpleMaps.com. Cities not included in that dataset were manually matched to county FIPS code. Incidents reported on university campuses were referenced to their respective counties. A small number of townships, villages, and towns could not be located, likely due to reporting errors, and were thus excluded.

The main analyses tested our hypotheses suggesting that the three measures of antisemitism either will (Hypothesis 1) or will not (Hypothesis 2) substantially correlate with one another. We did this by assessing the Spearman correlation between the three measures by county. For this analysis, the three AMCHA variables were summed to create an overall index of AMCHA incidents.

We used Spearman correlations because the data were not normally distributed; most counties had no antisemitic incidents. Consistent with Hypothesis 1, the three measures correlated substantially [Spearman correlations: r_{(3, 106)} = 0.56 (FBI with ADL); r_{(3, 106)} = 0.53 (ADL with AMCHA); and r_{(3, 106)} = 0.48 (FBI with AMCHA); all p's < 0.001].

Figure 1 shows three maps of the United States, each colored by the percentile of the count of antisemitic incidents occurring in each county per the titled dataset. The maps reflect: 1. The different overall levels of antisemitism reported by the three sources; and 2. The geographic similarity in the patterns of antisemitism each source reports. The first map shows the percentile rankings of counties based on the Targeting variable reported by the AMCHA Initiative (because it is most similar to the ADL and FBI data), while the second map displays the FBI's reported antisemitic hate crimes, and the third map represents the Anti-Defamation League (ADL) reports of antisemitic incidents. Each county's color intensity corresponds to its relative percentile ranking, with darker shades indicating higher frequencies or reports of antisemitic incidents.

The maps provide a geographical perspective on the prevalence and distribution of antisemitism at the county level, aiding in understanding regional patterns and differences. They are also a way to visualize the overlap or similarity among the three measures of antisemitism and are, therefore, another way to examine our alternative hypotheses. Taken together, the visual perspective of the maps and the correlations among the three data sources all support Hypothesis 1 and indicate that there exists substantial alignment on county-level antisemitism measures. Therefore, Studies V, VI, and VII use the data from the FBI, ADL, and AMCHA on antisemitism.

We examined whether levels of each of the three AMCHA types of incidents varied depending on the receipt or source of Section 117 funding received from 2014 to 2019 (see Supplementary Tables S2–S4).

Research Question 1: Are there higher levels of antisemitism/anti-Zionism at institutions that received Section 117 funding than at institutions that did not?

Four Welch's t-tests compared levels of BDS, Expression, Targeting and their summed total among institutions that either received or did not receive Section 117 funding. Table 7 reports the results. All three measures and the total were higher among institutions that received Section 117 funding (all p's < 0.001). Furthermore, the effect sizes were strong (0.52 ≤ Cohen's d's ≤ 0.74).

Research Question 2: Are there higher levels of antisemitism/anti-Zionism at institutions that received Section 117 funding from OIC member countries than at institutions that received Section 117 funding from other countries?

Four Welch's t-tests compared levels of BDS, Expression, Targeting and their summed Total, but this time, among institutions that received Section 117 funding from either OIC or other countries. Table 8 reports the results. These results were also strong and consistent. All three measures of antisemitism/anti-Zionism, and the total, were much higher among institutions that received OIC funding (all p's < 0.001, all d's ≥ 0.59; see Table 8).

Research Question 3: Are there higher levels of antisemitism/anti-Zionism at institutions that received Section 117 funding from authoritarian countries than institutions that received Section 117 funding from non-authoritarian countries?

Four Welch's t-tests compared levels of BDS, Expression, Targeting and their summed total, but this time, among institutions that received Section 117 funding from either authoritarian countries or other countries. Table 9 reports the results, which were also strong and consistent. All three measures of antisemitism/anti-Zionism, and the total, were much higher among institutions that received funding from authoritarian countries than from other countries (all p's < 0.001, all d's ≥ 0.55; see Table 9).

Overall, therefore, Study V produced strong evidence that receipt of Section 117 funding per se is associated with heightened antisemitic/anti-Zionist activity. It also produced consistently strong evidence that institutions that received Section 117 funding from OIC member countries or authoritarian countries had much higher levels of antisemitic/anti-Zionist activity.

This study examined whether there were relationships between AMCHA data on campus antisemitism/anti-Zionism and ADL and FBI data on county antisemitism, using a county-level dataset (Study VI used the same data from 2015 to 2020 that was used in Studies IV and V). The dataset included 3,108 counties within the contiguous United States. For this analysis, we used AMCHA Targeting (rather than AMCHA BDS or Expression) due to its similarity in kind to ADL-recorded incidents of antisemitism and FBI-reported hate crimes. County demographic data was obtained from the U.S. Census Bureau's (2017) American Community Survey. These data were then merged with a TigerLine county shapefile after removing states and territories outside of the lower-48 states using FIPS codes.

Study VII examined the relationship between campus antisemitism (AMCHA data, 2015–2020) and antisemitic hate crimes in the U.S. as reported by the FBI from 2015 to 2020. Because Study VII focuses on antisemitism per se, it did not include analyses with AMCHA's BDS data. Supplementary Table S5 summarizes the data on antisemitism and hate crimes against Jews. The FBI and AMCHA data were integrated into a daily time series table using INDEX and MATCH functions in Excel. ADL data were not used in these analyses because they are not available by year for 2015 to 2020.

Analyses examined whether variability in campus antisemitism forecasts variability in antisemitism in the U.S. Figure 2 visually displays the variability and covariability among FBI antisemitic hate crimes and antisemitic incidents on campus. Spikes in green or blue (AMCHA campus antisemitism) often precede spikes in FBI reported antisemitic hate crimes (red).

To test this pattern more rigorously, we employed Granger causality analysis. Granger causality is a statistical time-series correlation analysis which includes lagged variables to determine whether one is useful in forecasting the other. This assesses whether lagged values of the explanatory variable better predict the dependent variable than the latter's lagged values alone (Stern, 2019).

Granger causality has some important limitations. The term “Granger causality” is something of a misnomer because forecasting is not equivalent to causality and the results of a Granger causality analysis rarely prove causality (Shojaie and Fox, 2022). In addition, effects sizes are not produced by Granger causality analyses. Despite its limitations, Shojaie and Fox (2022, p. 314) conclude that: “At minimum, …continued developments in this area can help researchers take the first step toward causal inference by restricting the set of possible causal hypotheses.” See Supplementary Table S7 for fit measures.

The key Granger causality results are shown in Table 11. Both Campus Targeting Incidents (p < 0.0001) and Campus Antisemitic Expression (p = 0.0257) forecasted FBI-reported hate crimes. However, the p-value above 0.01 should be viewed as an invitation to future research rather than as definitive. The Granger causality results also found that national FBI hate crimes forecast Campus Antisemitic Targeting incidents (p = 0.0039), but a statistically significant forecast was not found for Campus Antisemitic Expression (p = 0.159).

These findings are exploratory and correlational, meaning that causal conclusions are not justified, and further research is necessary. Nonetheless, the Granger causality analyses indicated that both campus antisemitic targeting and FBI hate crimes nationwide forecast one another. Speculation regarding hypotheses worthy of further testing are, therefore, warranted. For example, one possible explanation for these relationships is that institutional antisemitism could spill out to national incidents and hate crimes. Another is that heightened antisemitism around the country could manifest on campus.

The strongest and clearest results indicate bidirectional predictive power between Campus Antisemitic Targeting Incidents and FBI hate crimes, whereas relationships with Campus Antisemitic Expression were weaker. We speculate that this pattern occurs because, like targeted antisemitic harassment, hate crimes (other than property-related crimes) target a specific person or people. Also, antisemitism directed at specific targets is more severe than is un-targeted expression. Harassment, bullying, or hate crimes (targeting) are clearly more severe transgressions than antisemitic or anti-Zionist posters or chants (expression), which is why targeting is potentially punishable either by legal or campus authorities whereas expression rarely is (a notable exception is when it constitutes vandalism or other property-related crimes). As such, targeting may be more likely to make campus, local, or even national news stories. If so, this might be more likely to inspire similar such actions in other places. Regardless of the explanation, however, our findings indicate that campus institutional antisemitism does not remain isolated to the university but is related to broader regional activity.

As the first large-scale and data-driven report that links foreign funding, democratic norms and antisemitism, this work has notable limitations. Because our analyses were correlational, we did not reach causal conclusions. This leaves unanswered many important questions. For example, does receipt of foreign funding, especially from member countries of the Organization for Islamic Cooperation or from authoritarian countries, increase antisemitism, illiberalism or the erosion of specific liberal democratic norms on campus such as those regarding free speech? Or perhaps the independent development of illiberalism and antisemitism attracted such funding. Does some third variable (such as university status, number of state-sponsored students from anti-democratic countries, number of antisemitic faculty, whether the university president is an antisemite, etc.) cause illiberalism and attract foreign funding from illiberal sources? Do spikes in campus antisemitism cause spikes in surrounding communities and the country? Do spikes in local or national antisemitism cause spikes in campus antisemitism? Or does some combination of causes combine in complex ways?

This research provided no insight into how any of the potential directions of causality might occur. Perhaps Section 117 money funds illiberal or antisemitic groups on campus. Perhaps it supports hiring illiberal or antisemitic professors. Perhaps this sort of support provides tacit endorsement of antisemitic incidents. Alternatively, campus sociopolitical trends conducive to illiberalism and antisemitism may have developed independently of Section 117 funding, and Section 117 funding may have no causal relationship whatsoever with campus illiberalism or antisemitism.

Another limitation stems from the exploratory nature of all studies reported herein. Our analyses cast a broad net, operationalizing “authoritarian sources” as the 30 most authoritarian countries plus Russia. Separately, for OIC countries, most were ranked as authoritarian or “hybrid” (a mix of democratic and authoritarian) by the Democracy Index, and many have populations with high levels of antizionist and antisemitic beliefs and attitudes. One could test these hypotheses using different operationalizations of authoritarianism and antisemitic/antizionist attitudes. We did not do this because it could have muddied the findings, due to overlap between various alternative operationalizations. Additionally, different operationalizations would have created a massive multiple comparisons problem requiring a dramatic reduction of our already-reduced threshold for statistical significance. Although, the consistent differences found between these and other countries validates how we grouped them, it is nonetheless possible that those differences derive primarily from specific subsets of those countries we grouped. Addressing these sorts of issues is an important issue for future research.

Future research should test potential causal explanations. Whereas experimental research, which would permit strong causal conclusions, is probably not possible in this area, identifying causal directions is an important area for future research. Causal hypotheses could be tested in future non-experimental research that might justify causal conclusions, as long as certain strong standards are met (Rohrer, 2018). Although the present research does not meet those standards, it does raise important causal questions.

Additionally, the present research did not assess why some of this funding went unreported for years, until the U.S. Department of Education conducted an investigation in 2019. Therefore, the present research did not examine whether this occurred because of innocent mistakes, managerial incompetence, overly complex bureaucratic reporting requirements, political agendas, corruption, or some combination of reasons. It is also possible that foreign funding went unreported for different reasons at different institutions. The purpose of the present research was to investigate social and political phenomena related to receipt of Section 117 funding; its purpose was not to investigate how or why the money went unreported, though future research doing so would be valuable.

Last, this was the first quantitative investigation (of which we are aware) on how receipt of Section 117 funds relates to campus liberal democratic norms. As such, it was exploratory, rather than confirmatory. Therefore, all findings should be viewed as preliminary, pending replication in future research.

It is important not to misunderstand or misrepresent the findings in the present paper. Many elements of the findings may appear ripe for conspiratorial thinking—secrecy (undisclosed funding), wealthy and powerful foreign actors, and harmful outcomes (e.g., spreading antisemitism and undermining liberal democratic norms). Nothing presented here suggests that mustachio-twirling villains from authoritarian or majority Muslim countries are conspiring in dark basements with deans and provosts to corrode democracy or harm Jews. We caution against misconstruing these findings to blame and target students or faculty who simply have ties to these donor countries. Indeed, we have no evidence that any of the relationships we uncovered reflect causal effects of foreign funding. Causal effects are one possible explanation for our findings, but, as we have pointed out throughout the manuscript, causality may flow in very different directions. Furthermore, the main outcomes in the present study—liberal democratic norms around speech and academic freedom, and antisemitism, are surely influenced by many factors not addressed in any of our studies. Moreover, even if the influence of foreign funding is ultimately found to be causal in subsequent research, it would likely happen through very conventional processes (e.g., attracting or supporting scholars or organizations with particular views or political agendas) rather than through conspiracies concocted in dark corners.