Buffeted by Change, Researchers Seek Continuity and Investment

Executive summary

The research enterprise in the United States, established after the second world war, is experiencing massive disruption. US colleges and universities are grappling with federal and state disinvestment and intervention at a time when AI and other technologies are reshaping how research is conducted and its findings are communicated. As a result, researchers at US higher education institutions are operating in an unstable environment. Against this backdrop, the findings from our survey of 4,003 researchers, fielded in late 2025, demonstrate that they remain committed to carrying out essential work and that they are doing so in innovative ways—through collaboration, the creation of valuable educational experiences, and pushing knowledge towards the next frontier. While researchers don’t necessarily agree on how their sector should respond to the changing environment, our findings underscore the importance of stability, continuity, and sustained investment.

Key findings

• AI is producing deep divisions among researchers, but disclosure has emerged as a shared ethical baseline. Researchers distribute responsibility for governing AI use across the scholarly ecosystem, a model that could support collective stewardship but may also leave critical accountability gaps.
• While access to resources varies widely, government funding is critical to the research enterprise. Institutions have limited capacity for a sustained federal retreat.
• Despite resource restraints, researchers generally feel valued and supported by their institutions, with libraries emerging as a particularly important source of stability.
• Collaboration knits together the research enterprise, creating connections through which ideas circulate and discoveries are made. Students are important participants in these networks, underscoring how closely research and teaching are intertwined.
• The peer-reviewed article remains the coin of the realm, but it represents only the most visible part of a much richer scholarly ecosystem. Researchers produce work for students, communities, and other audiences while contributing the often-unrecognized labor—mentoring, peer review, research assistance, and leadership—that makes knowledge production possible.

Introduction

The American research enterprise—that is, the people, the resources, and the ideas that support a system of transformative innovations that benefits US society and the global community—emerged from the fulcrum of World War II under the leadership of Vannevar Bush. During the war, federal funds led to the development of critical technologies such as radar and the mass production of penicillin, a valuable proof of concept of what scientists could accomplish with federal support and of the importance of a vibrant research enterprise to national security.^[1] The resulting partnership between the federal government and universities—where scholars in higher education receive federal funding to conduct research for the benefit of all—became firmly institutionalized in the decades after the war through agencies like the National Science Foundation (NSF), an expanded National Institutes of Health (NIH), and the National Aeronautics and Space Administration (NASA). This partnership has fueled American leadership in scientific discovery for the better part of a century.

The spread of generative AI as a mass-market technology is rapidly changing research and scholarly communication practices across disciplines.

In the 2020s, the US academy has been rocked by changes which have destabilized this system. The spread of generative AI as a mass-market technology is rapidly changing research and scholarly communication practices across disciplines. Executive actions and the cancellation of thousands of federal research grants—as well as corresponding attempts to close agencies such as the National Endowment for the Humanities (NEH) and the Institute of Museum and Library Services (IMLS) and cut the budgets of others, such as the NSF, by over 50 percent—are creating massive uncertainty about the future of federal funding for basic research.

The federal government along with a number of states are also asserting new levels of control over what topics can be researched at public institutions or with government funds. In our first report of findings from the 2025 US Researcher Survey, published in April 2026, we assessed the impact of state and federal laws and policies on research practices. We found that 20 percent of the nation’s researchers have avoided certain topics or shifted their research agenda because of these new laws and policies and that researchers have deep concerns about the future of federal support for the research enterprise.^[2] We do not repeat our analysis of that section of the survey in this report, but it’s important context for our findings.

In this time of volatility and uncertainty, Ithaka S+R sought to assess the status of the research enterprise by surveying the researchers who sustain it. Since 2000, we have used national faculty surveys to provide US colleges and universities, libraries, learned societies, and academic publishers with insight into the evolving attitudes and practices of faculty members so that they can better support faculty needs. Our findings speak directly to the uncertain funding environment, the impact of AI on research integrity, and a range of other critical issues, providing insights that will help librarians, research offices, and scholars support and sustain the research enterprise during a period of systemic transition.

Methodology

The survey sample consisted of a list of 135,532 higher education contacts purchased from a vendor. The sample included people working at four-year nonprofit institutions who had recent experience as instructors of record, although it also included some deans and administrators. For each person in the sample, the vendor provided a discipline keycode and a department code (based on courses taught). The survey was administered through Qualtrics between September 22 and November 17, 2025. Respondents who did not affirm that they conducted “research, scholarship, or creative activity” were disqualified.

The cleaned analysis sample included 4,003 complete and partial responses from researchers at 789 unique institutions, with a median of three respondents per institution and an overall response rate of 3 percent.^[3] Most respondents worked at public institutions (66 percent), with 34 percent at private not-for-profit institutions. Sixty-four percent of respondents worked at doctoral institutions, 22 percent at master’s institutions, 11 percent at baccalaureate institutions, and 3 percent at special focus institutions, which in our sample were mostly medical schools.^[4] Respondents were fairly evenly distributed among the four Census regions, with 33 percent working at institutions in the South, 25 percent in the Northeast, 23 percent in the Midwest, and 19 percent in the West. Respondents were also fairly evenly distributed by institution size: 24 percent worked at small institutions (5,000 students or less); 30 percent at medium institutions (5,001-16,000), 16 percent at large institutions (16,001-25,000), and 30 percent at very large institutions (more than 25,000).

Figure 1 – Percentage of respondents by discipline^[5]

We designed our sample to include broad definitions of research and of institutional affiliation. Recognizing that research, scholarship, and creative activity take place at institutions of all types, and among people with varying relationships to their institutions, we aimed to study how variation in researchers’ access to resources (such as time, money, and institutional support) impacts the research enterprise as a whole.

More than a quarter of respondents (27 percent) are not tenure-line faculty. Their roles include full-time non-tenure track faculty (12 percent), part-time faculty (5 percent), advanced graduate students (3 percent), retired faculty (2 percent), staff or administrators (2 percent), and other roles (3 percent). As one respondent wrote, “Thanks for recognizing the diversity of ways that people can be employed in teaching and research. As an adjunct at several institutions who invests his own time and resources to engage in research and support the discipline, I often feel left out of consideration.” Most respondents’ duties extend beyond research: 94 percent teach, 80 percent spend time on institutional and other forms of professional service, 33 percent have administrative responsibilities, and 4 percent provide clinical services.

“As an adjunct at several institutions who invests his own time and resources to engage in research and support the discipline, I often feel left out of consideration.”

Researchers agree on principles of research integrity, but generative AI complicates their application

Concerns about the integrity of the scholarly record have been growing for some time. Academic incentive structures, not least the pressure scholars face to publish frequently, are often seen as a key factor driving research misconduct. Among our respondents, 53 percent agreed or somewhat agreed that this was the case. These incentives have not changed since the commercialization of generative AI, but the technology has dramatically lowered barriers to misconduct, especially with respect to the publishing process. A growing body of literature suggests that many scholars are using the tool to help them write journal articles without disclosure, creating new threats to the integrity of the scholarly record.^[6]

Our survey included several questions designed to better understand researchers’ assessment of the extent of these problems and beliefs about how responsibility for policing them should be distributed. Our findings indicate considerable areas of agreement about which forms of misconduct are particularly egregious and highlight shared understandings of important principles of research integrity.

Figure 2 – Which five of the following do you consider to be the most severe forms of research misconduct?
Percentage of respondents who selected each option; respondents were required to select five.

Three research practices were widely considered to be among the most severe violations of research ethics: the misrepresentation of findings (84 percent), plagiarism (82 percent), and citing false sources (71 percent), indicating that truthfulness and proper attribution are core principles shared across disciplines. A majority of respondents listed claiming fraudulent credentials, another form of misattribution, as a major ethical lapse. These types of misconduct occur during the publication process, but two other widely cited forms of misconduct, the collection of data without consent (47 percent) and inappropriate use of generative AI to conduct research (58 percent), focus on integrity issues that emerge during the research and experimentation process.

Generative AI creates disagreements about how shared principles should be rooted into research practices and oversight. Most importantly, respondents are divided about whether AI use during the research lifecycle is ever appropriate. Reported use of AI for research purposes varied widely by discipline, with non-users concentrated in the arts, humanities, and social sciences.

Figure 3 – Generative AI is defined as Artificial Intelligence models that have been trained on existing data and can generate content from that data. At what stage of conducting research, scholarship, or creative activity are you most likely to use generative AI? (Select one only.)
Percentage of respondents who selected the option “I don’t use generative AI while conducting research, scholarship, or creative activity,” by discipline.

Overall, 50 percent of respondents reported that they do not use AI for research purposes, and open-ended responses show strong objections to the technology among some researchers. “I find AI one hundred percent threatening to nearly every value that I hold in teaching, research, and writing,” wrote one interdisciplinary studies scholar. A social scientist wrote, “AI poses an existential threat to both humanity and the academy.” According to a communications scholar, “Generative AI is a disaster for scholarship on every level.” Extreme hostility towards AI is a minority opinion, but by no means a fringe view. Overall, 32 percent agreed with the statement that generative AI should not be used for scholarly purposes, while 46 percent disagreed, and 22 percent weren’t sure.

Thirty-two percent agreed with the statement that generative AI should not be used for scholarly purposes, while 46 percent disagreed, and 22 percent weren’t sure.

Whether generative AI should be used in research is contentious, but our respondents expressed a strong consensus on other principles of research ethics. The largest area of agreement focused on the importance of disclosure: 86 percent of respondents agreed or somewhat agreed that whether AI was used must be disclosed, and 88 percent agreed or strongly agreed that researchers must disclose how they used it. Two-thirds of respondents voiced opposition to the use of generative AI by peer reviewers.

In contrast to the high levels of concern over the use of AI by reviewers, researchers appear broadly open to its use by scholarly publishers: just 26 percent of researchers believe that publishers should not use generative AI for any purpose. As a communications scholar wrote, “I don’t mind if publishers use generative AI or AI more generally as long as there’s a person overseeing the process.” Researchers were particularly likely to support the use of generative AI to detect research fraud; as one chemist noted, “if sufficient protections for the data submitted to a generative AI platform were in place, it may be acceptable for publishers to use generative AI to vet work.” Almost three-quarters (71 percent) said that publishers should be able to use it to identify plagiarism. About two-thirds agreed that publishers should be free to use generative AI to detect falsification of images (69 percent) and statistics (67 percent).

Figure 4 – Please indicate the degree to which you agree or disagree with the following statements.
Percentage of respondents who selected each option.

Figure 5 – Please select all of the following statements with which you agree.
Percentage of respondents who selected each option.

While researchers clearly see publishers as having an important role in maintaining the integrity of the scholarly record, they see them as only one among many entities with a role to play in regulating generative AI usage in the research enterprise. Researchers identified this responsibility as spread widely across the scholarly ecosystem: in aggregate they believed publishers bore 24 percent of this responsibility, followed closely by academic institutions (23 percent), and disciplinary communities (17 percent). Funders and the government were seen as having a lesser role.

Figure 6 – In your view, what percentage of the responsibility to regulate the use of generative AI for research, scholarship, or creative activity does each of the following entities bear? Please type a number between 0 and 100 in each box. Numbers should sum to 100. Mean of respondents’ answers for each option.

Research is under threat across academic ranks and institution types

Research is a resource intensive endeavor. We asked respondents a series of questions designed to help us better understand how access to critical resources, especially time and money, is distributed across the research enterprise. That distribution is, of course, uneven. Respondents in STEM fields, who are more likely to have access to federal research funding, had greater access to both time and money than respondents in other fields. Institutional funding has historically been distributed more equitably across disciplines, but uncertainty about future funding for important federal agencies could force institutions to consider concentrating their resources to support research in fields they identify as strategically important.

Federal funding is critical to the research enterprise

For the past half-century, the majority of US academic research dollars have come from the federal government.^[7] Nearly half of our respondents (43 percent) reported receiving some level of research funding from the government within the past five years. The vast majority of those funds were distributed by federal agencies: the federal government provides roughly 53 percent of academic research funds annually, while state appropriations account for just 5 percent.

Figure 7 – In the past five years, approximately what percentage of the total funding you have received or are currently receiving comes from each of the following sources? – External funding from a governmental grant-making organization (such as the NSF, NIH, NEH, NEA, etc.).
Mean percentage of funds from government sources by total funds received over the past five years.

These funds are unevenly distributed by discipline and by Carnegie classification; unsurprisingly, respondents at doctoral institutions and respondents in STEM and the health sciences disciplines were more likely to report receiving this type of funding.

Figure 8 – In the past five years, approximately what percentage of the total funding you have received or are currently receiving comes from each of the following sources?
Percentage of respondents who reported receiving each type of funding, by Carnegie classification.

Figure 9 – In the past five years, approximately what percentage of the total funding you have received or are currently receiving comes from each of the following sources?
Percentage of respondents who reported receiving each type of funding, by discipline.

Universities themselves are the second largest source of academic research dollars nationally. Institutional funds are particularly critical for supporting research in fields without ready access to extramural funding but are a resource to researchers across fields. Our findings suggest that internal funds are distributed more widely than federal funds: 73 percent of our respondents reported receiving funding from their home institution to conduct research.^[8] Researchers at master’s and baccalaureate institutions and those who reported receiving $10,000 or less over the past five years were particularly likely to have received internal funding. If the current slowdown in federal funding continues or deepens, institutions will likely face difficult choices about whether and how to redistribute this source of funding. For researchers in fields with few other options, this could lead to increased reliance on personal funds to subsidize their scholarly activities.

Figure 10 – In the past five years, approximately what percentage of the total funding you have received or are currently receiving comes from each of the following sources?
Percentage of respondents who reported receiving each type of funding.

The amount of monetary support available to researchers varies widely, ranging from zero to tens of millions of dollars. Among respondents who reported receiving any amount of research funding during the past five years, 66 percent received less than $100,000. Unsurprisingly, more research funding was available to respondents in STEM and health sciences fields (especially those at doctoral institutions), where research costs are typically more substantial and funding opportunities are more readily available. Much of that funding comes from federal agencies such as the NIH, NASA, and NSF, where the administration is seeking budget cuts of 12, 24, and 52 percent, respectively.^[9] While Congress has blocked these proposed cuts from taking effect, they still create uncertainty for researchers in STEM and health sciences fields.

Figure 11 – In the past five years, how much funding in total have you received or are you currently receiving to fund your research, scholarship, or creative activity?
Percentage of respondents who reported receiving funding in each category.

Time is researchers’ most precious resource

While sufficient funding is critical, perhaps the scarcest resource available to researchers, regardless of discipline or institution type, is time to pursue their research agenda. “Time is the biggest constraint,” noted one respondent. Another remarked that money made little difference if “there is no time allocated for research work.”

Six percent of our survey population work as full-time researchers, 13 percent have no contractual commitment to conduct research, and the majority fall somewhere in between. Consistent with the high percentage of tenure-line respondents, the median respondent was expected to devote 33 percent of their time to research, scholarship, or creative activity. Regardless of their institution’s size or their job role, researchers with more time contractually dedicated to research were more likely to report feeling that their institution valued their research; those who were most dissatisfied with their institution’s degree of research support contractually devoted on average 25 percent of their time to research, while those who were most satisfied devoted 40 percent of their time.

Figure 12 – Please indicate the degree to which you agree or disagree with the following statements: My institution values my research, scholarship, or creative activity.
Percentage of respondents who selected each option by how much time they had contractually devoted to research.

Researchers in STEM fields and especially in medicine had a higher median percentage of their time contractually devoted to research than researchers in other disciplines.

Figure 13 – In a typical year, approximately what percentage of the time you devote to professional activities is contractually devoted to research, scholarship, or creative activity?
Median percentage of contractual research time by discipline.

Research time was also highly dependent on Carnegie classification. As expected, median time contractually devoted to research was higher (40 percent) at doctoral institutions, the small number of universities that have made research central to their missions. Researchers at teaching-oriented institutions understandably had less contractual time devoted to research: at master’s institutions it was 20 percent and at baccalaureate institutions it was 25 percent. Within different types of doctoral and baccalaureate institutions, time budgets also varied dramatically. Researchers at special focus institutions, primarily medical schools in our sample, had the most research time.

Figure 14 – In a typical year, approximately what percentage of the time you devote to professional activities is contractually devoted to research, scholarship, or creative activity?
Median percentage of contractual research time by Carnegie classification.

Despite the challenging resource environment, researchers feel reasonably well supported by their institutions

Researchers rely on the support services available on their campus to make research possible, and our survey responses clearly highlight the library as a core campus resource that supports the research enterprise. Researchers identify the library as their institution’s primary provider of research support and view the quality of the library’s service provision very favorably.

Libraries are the mediators of access to the scholarly record, which 86 percent of respondents considered to be among the top five resources provided by their institution. No other form of research support attracted as much consensus as library collections, a reflection of their position at the center of scholarly inquiry. While many other types of support services offered by institutions cater to the needs of some fields more than others, collections are uniquely central to researchers across fields. Twenty percent of respondents considered datasets, another form of collection that is often accessed through libraries, as a top five resource, and 26 percent placed reference services as a highly valuable research support.

Fifty-eight percent of respondents named the library as an important provider of research support services, second only to their academic unit.

In recent years, libraries have actively worked to highlight that their role extends beyond collections to service provision.^[10] While our findings highlight the enduring value of collections to researchers, the resources that the library has devoted to service provision are also rated highly by many researchers. Fifty-eight percent of respondents named the library as an important provider of research support services, second only to their academic unit (76 percent). Researchers at master’s and baccalaureate institutions are particularly reliant on libraries for research support services, a finding consistent with our previous research documenting the centrality of the library to research support services, and especially at non-doctoral institutions.^[11]

Several services provided by research offices, including assistance with grant proposals (48 percent), access to core facilities (51 percent), and access to high-performance computing and data storage were also rated highly by many researchers. Twenty-eight percent cited IT or research computing as an important provider, while 16 percent cited their local research institute or center. Respondents more rarely included resources such as the IRB, legal services, or the tech transfer office in their top five resources.

Figure 15 – Which of the following resources are most valuable for your research, scholarship, and creative activity? Select up to five resources.
Percentage of respondents who selected each option.

Figure 16 – For each resource supporting your research, scholarship, or creative activity, who is the primary provider? Please select up to 5 choices.
Percentage of respondents who selected each option.

In addition to reporting which campus resources they used, respondents also rated how effectively different campus units supported their research. Respondents were most satisfied with the support they received from their libraries (68 percent somewhat agree or agree). Respondents were also generally satisfied with the degree to which their institution as a whole valued research (64 percent), and with their institution’s assessment of their research in making decisions about retention, tenure, and promotion (61 percent). Fifty-eight percent of respondents expressed satisfaction with research computing and core facilities.

Figure 17 – Please indicate the degree to which you agree or disagree with the following statements. Percentage of respondents who selected each option.

Collaboration is ubiquitous

Collaboration, which we define as individuals working together to achieve a common goal, is a hallmark of contemporary research, and researchers report collaborating with people in a variety of roles. Students—including undergraduates—are important research collaborators.

Respondents were most likely to collaborate with people who were close at hand—typically people at their institution or within the United States. Researchers most frequently collaborate with faculty: 75 percent of respondents said they collaborated with faculty at their institution, and 92 percent of respondents said they collaborated with any faculty member. Undergraduates also have an important footprint in the research enterprise: 60 percent of respondents at doctoral institutions, 57 percent of respondents at master’s institutions, and 71 percent of respondents at baccalaureate institutions stated that they collaborated on research with undergraduates.

The high number of researchers who report collaborating with undergraduates is one of several reminders of the permeable barrier between teaching and research that our survey uncovered.

Much of this collaboration is primarily oriented towards promoting student learning.^[12] However, the high number of researchers who report that they collaborate with undergraduates is one of several reminders of the permeable barrier between teaching and research that our survey uncovered. The literature on undergraduate contributions to the creation of knowledge is small, but clear. Undergraduates are involved in many core research activities such as data collection and curation.^[13] They are also co-authors of a surprising number of publications: a bibliometric study of medical literature found undergraduate co-authors listed on 10 percent of journal articles, while a survey of psychology faculty found that 83 percent had published at least one peer-reviewed article with an undergraduate co-author and over half had published peer-reviewed material with an undergraduate as a first author.^[14] Our findings suggest the need for further investigation of faculty and student collaboration as a direct contribution to the research enterprise.

Likely buoyed by large grant programs such as the NSF’s Research Experiences for Undergraduates program—which is among the NSF programs targeted for budget cuts by the administration—respondents in STEM fields were most likely to collaborate with undergraduates at their institution, with 80 percent doing so. But collaboration with undergraduates cuts across disciplines: 61 percent of our respondents in the humanities and social sciences and 56 percent of those in the fine arts reported collaborating on research and creative activities with undergraduates.^[15]

Figure 18 – Collaboration is defined as individuals working together to achieve a common goal. In the past five years, have you collaborated with any of the following people to conduct research, scholarship, or creative activity? For each category, please note whether you collaborated with people at your institution or another institution, in the US or outside the US, and in your discipline or another discipline. Select all that apply.
Percentage of respondents who selected each option shown, by discipline.

Despite the ubiquity of collaboration, the survey findings suggest that researchers would be open to even more collaboration if they had the resources to do so. While respondents listed their primary barriers to collaboration as limitations around time (74 percent) and money (57 percent), 32 percent cited difficulty identifying collaborators as a barrier, and 29 percent cited administrative difficulties.

Figure 19 – Which of the following factors serve as the biggest barriers to collaboration for your research, scholarship, or creative activity? Select up to five barriers.
Percentage of respondents who selected each option.

The article is still the coin of the realm

While important efforts have been made to improve the status of a range of scholarly outputs, actual scholarly practices remain fairly traditional. Conference presentations and journal articles were easily the most common methods researchers used to share research findings. Eighty percent of respondents presented their work at conferences or similar venues within the past five years, and 79 percent published at least one peer-reviewed article. Articles and presentations were the most common scholarly output in nearly every discipline for our survey population, with two partial exceptions. In the fine arts, exhibitions or performances were more common than journal articles (59 percent and 44 percent, respectively), and in computer science 80 percent of respondents had published in conference proceedings compared to 72 percent who had published peer-reviewed articles or the 62 percent who had given a presentation of their work.

Eighty percent of respondents presented their work at conferences or similar venues within the past five years, and 79 percent published at least one peer-reviewed article.

Disciplinary differences are most evident in the stark contrast between “book fields” and non-book fields. Respondents in STEM and health sciences fields were far less likely (11 and 8 percent respectively) to have published a monograph or in a collection of essays within the past five years, for example, while in the humanities and social sciences, these outputs were a common feature of the scholarly communication landscape (51 percent languages and literatures; 40 percent humanities and social sciences). Besides books, a few other research outputs are field-specific: preprints, for example, are important parts of the landscape in many STEM disciplines (26 percent) and have little presence in others (e.g., 7 percent in the professions; 3 percent in languages and literatures).

For the most part, however, the most common types of research outputs are relatively consistent across disciplines. Despite considerable efforts by funders and other advocates of open science, data publication is uncommon across most fields. Fewer than three in 10 researchers in agriculture and the natural sciences, and two in 10 computer scientists or medical researchers, listed this output as one they had produced over the past five years. Few respondents outside computer science, mathematics, or engineering reported sharing code or software. Our survey is almost certainly underestimating the prevalence of data and code sharing, which is often done through indirect channels that researchers might hesitate to define as a scholarly output, such as by sharing directly with a colleague.^[16] Nevertheless, our findings underscore the challenges of recognizing them as scholarship.

In contrast, community engagement and pedagogically oriented materials are common in most academic fields. Community engagement was most important in nursing, where 47 percent of respondents reported involvement, but an average of 21 percent of scholars across disciplines have been involved in community engagement within the past five years. Educational outputs were another cross-cutting research output: 20 percent of respondents across disciplines reported publishing textbooks, reference books, or other pedagogical materials.

Figure 20 – Consider how you have shared the results of your research, scholarship, or creative activity during the past five years. Please select up to five types of scholarly outputs that you have produced most frequently during that time. Percentage of respondents who selected the top six options in each discipline.

Figure 21 – Suppose you are evaluating a colleague in your discipline for retention, tenure, or promotion at your institution. Your colleague has produced the following outputs of research, scholarship, or creative activity. Please select the five types of scholarly outputs that you believe to be most impactful. Percentage of respondents who selected the top six options in each discipline.

We also asked researchers about how they assess the value of different scholarly outputs in tenure and promotion review. Their responses underscore the centrality of the article to scholarly communication: 93 percent of all respondents included it among their top five outputs. Strong majorities of respondents highly valued scholarly monographs and textbooks or other educational materials (60 percent and 59 percent, respectively). While 80 percent of researchers reported that they frequently give presentations, only 50 percent named them as highly impactful when evaluating their colleagues.

Figure 22 – Comparison of top six academic outputs by whether it was produced by the respondent or a hypothetical peer.
Percentage of respondents who selected each option; highest number in each row highlighted.

Service in support of the research enterprise is distributed broadly

In addition to producing research outputs, researchers also perform a range of professional activities that keep the research enterprise functioning. The most common of these, according to our respondents, are serving as peer reviewers (54 percent of respondents do this very often or always) to facilitate scholarly communication, helping other scholars with research (39 percent), supervising graduate students (46 percent) to train the next generation of researchers, and serving as a principal investigator (PI) (45 percent) to manage collaborative research labor. While some of these activities are common across all disciplines, in other cases disciplines tend to specialize, each playing their part to support the enterprise as a whole. For example, respondents in languages and literatures were more likely to serve as editors of a publication and to organize a conference; respondents in health sciences were more likely to be peer reviewers, provide research assistance to other scholars, and chair research committees; respondents in STEM and health sciences were more likely to serve as PIs, engage in research fundraising, and mentor junior researchers.

Figure 23 – Scholars make many contributions to the research enterprise in addition to outputs of research, scholarship, and creative activity. Over the past five years, how frequently have you done each of the following? Percentage of respondents who selected each option.

Academic rank also affected the type of contributions researchers make to the research enterprise; for example, full professors and retired faculty were more likely to edit publications and to serve as peer reviewers or committee chairs than other respondents. Tenured or tenure-track faculty were also more likely to engage in any specific research-related professional activities than their non-tenure track peers (mean of 50 percent vs. 37 percent across the 14 categories). Some of these activities are associated with levels of professional prestige that accumulate disproportionately to tenure-line faculty. At many institutions, tenure-line faculty have responsibilities, such as serving on committees or supervising graduate students, that are not shared by non-tenure-track faculty. Even so, our findings demonstrate that the work of supporting the research enterprise is distributed broadly among researchers regardless of their employment conditions.

Our findings demonstrate that the work of supporting the research enterprise is distributed broadly among researchers regardless of their employment conditions.

Figure 24 – Scholars make many contributions to the research enterprise in addition to outputs of research, scholarship, and creative activity. Over the past five years, how frequently have you done each of the following?
Percentage of respondents who selected each option at least sometimes, by role.

Researchers value curiosity, while their institutions value productivity

Researchers’ choices about which research to pursue are frequently based on their own personal interests and beliefs: 80 percent of researchers identified personal interest as motivating their work. However, several additional considerations play an important role in why scholars engage in research and the impact they hope their work will have. Scholars often identify themselves as members of specific disciplinary communities and describe their research as contributions to a field: 53 percent described the intended impact of their research as changing the views of other scholars in their discipline, and 44 percent hoped to gain the respect of their peers. The desire for career advancement (51 percent) and a belief that their research made them a better teacher (56 percent) were also common motivating factors. Broadly speaking, researchers’ primary motivations were fairly consistent across Carnegie classifications.

80 percent of researchers identified personal interest as motivating their work.

Figure 25 – Please select the top five motivating factors to you personally when you assess the impact of your research, scholarship, or creative activity.
Percentage of respondents who selected each option, by Carnegie classification; respondents were required to select five.

In contrast, researchers’ perceptions of their institutions’ priorities differed in important ways across Carnegie classifications, reflecting differences in institutional mission. Researchers at doctoral institutions were considerably more likely (73 percent) to identify grant funding as one of their institution’s top five priorities for assessing research impact. As one psychologist at an R1 wrote, “The only thing that my institution cares about in terms of deciding on pay raises is grants. Literally nothing else matters.” Meanwhile, at baccalaureate institutions, relevance to teaching and undergraduate involvement was the most frequently selected institutional priority (87 percent). At master’s institutions, researchers identified strategic alignment as a high priority (73 percent).

One criterion consistently identified across Carnegie classifications as a key institutional metric for research impact was the number of outputs, including publications, performances, exhibitions, and other scholarly products (78 percent doctoral, 67 percent master’s, 73 percent baccalaureate). Researchers everywhere believe this metric is central to how their institutions assess their research.

Elsewhere, differences in institutional mission are reflected in how researchers believe they are assessed. The sharpest distinction is how institutions weigh the intersection of scholarship and teaching. At baccalaureate and master’s institutions the relevance of research to teaching was the most frequently cited institutional priority for assessing research impact, whereas fewer than half of researchers at doctoral institutions included it among their top five. Researchers at doctoral institutions were much more likely than their peers at other institution types to report that impact factors and grant funding were top institutional metrics.

Figure 26 – How do you understand your institution’s priorities when it comes to assessing the impact of your research, scholarship, or creative activity? Please select your institution’s top five priorities.
Percentage of respondents who selected each option, by Carnegie classification; respondents were required to select five.

The relatively low frequency with which researchers identified media attention as an institutional priority suggests that public facing forms of impact are perceived as less central to research impact than traditional scholarly metrics such as publications and grants. This finding is notable given increasing pressure on universities to demonstrate the public value of research. Communicating the meaning and impact of scholarship to audiences beyond disciplinary peers requires skills and incentives that differ from those associated with scholarly publication. While researchers are often trained to communicate within disciplinary communities, there are far fewer perceived incentives and training for broader public engagement.

Conclusion

In retrospect, 2025 may prove to be a significant year for higher education, the point at which several converging pressures on the sector became difficult to ignore. The unforeseen withholding of research funds to individual grantees and entire institutions, cancellation of thousands of active grants, and a host of associated policies raise serious questions about whether the partnership between the federal government and higher education, a partnership that fueled the rise of the research universities that have sustained scholarly inquiry for decades, will continue.

Together, these forces are changing how researchers conduct their work, as well as injecting professional and institutional risk into the choices scholars make about what to study, forcing a reckoning with the very nature of what it means to learn and discover.

These disruptions are occurring simultaneously with the equally profound disruptions triggered by generative AI. Together, these forces are changing how researchers conduct their work, as well as injecting professional and institutional risk into the choices scholars make about what to study, forcing a reckoning with the very nature of what it means to learn and discover. Our survey of researchers was conducted at an optimal time for a snapshot of this tumultuous year: they capture researchers’ opinions in medias res but late enough in the year to allow for reflection and perspective about the state and future of academic research.

Academic researchers are spread across the landscape of higher education, though the time and resources available to them vary widely across employment conditions, institutions, and disciplines. Within this landscape, federal funding has played a critical role in supporting the research of 43 percent of all researchers over the past five years. Institutional funding is spread widely to support a much higher percentage of researchers. These are the largest sources of funding that fuel the research enterprise, and with federal funding at risk, institutions are facing difficult decisions about how to redistribute the resources at their disposal. The resulting layoffs, budget cuts, and shrinking of doctoral programs are early indicators of the limited capacity of even the best resourced universities to maintain the enterprise should the federal retreat from the research enterprise continue.^[17]

Our findings document areas of strength and stability within the research enterprise as well. Researchers are satisfied with the services and support provided by offices across their home institutions, and especially value the resources and expertise provided by university libraries. Solid majorities of scholars believe that their institution values their research and that promotion, retention, and tenure standards reflect that valuation. Despite rising geopolitical tensions, collaborative research appears to be thriving, creating points of contact between faculty, staff, and students that facilitate the flow of ideas and discovery.

Another point of stability is the continuing hegemony of the peer-reviewed article as the apex of scholarly outputs. The article shows no signs of being displaced, but as our findings demonstrate, it is surrounded by a host of other genres of scholarly communication. The less visible, but essential labor of mentoring, peer review, and conference organizing that researchers contribute to their fields are essential to the research enterprise.

Generative AI has created deep schisms between enthusiasts who are pioneering new ways of conducting and communicating research and a strong contingent of researchers whose principled objections to the technology must be taken seriously. Despite these divisions, disclosure has emerged as an important area of consensus on the basic ethics of AI use in research. There is less clarity about how to enforce those ethics. Researchers see AI as an important tool that can help publishers identify research misconduct before it enters the scholarly record, but they ultimately see publishers as only one node in a network spanning the research enterprise that bears collective responsibility for the veracity and integrity of academic research. Whether that dispersed responsibility portends a collective action problem or a communal solution will be worth tracking as AI becomes increasingly more pervasive.

Recommendations

Publishers

Establish clear standards for AI disclosure. Publishers should reach consensus on requirements for authors to specify whether and how generative AI contributed to research outputs without foreclosing individual journals from implementing additional requirements tailored to their specific disciplines and methodologies.

Implement AI-based integrity tools transparently and with human oversight. Publishers should regularly evaluate the accuracy and potential biases of tools used to detect plagiarism, image manipulation, and statistical anomalies. Information regarding these tools and their governance should be made publicly accessible. Humans should make final decisions regarding individual manuscripts.

Safeguard the integrity of the peer review process. Publishers should prohibit reviewers from submitting unpublished manuscripts to public AI systems, clearly delineate permitted uses of generative AI by reviewers, and disclose their own use of generative AI in editorial and production workflows.

Libraries

Adopt collective strategies to maintain access amid contracting collections budgets. Libraries should negotiate with publishers through consortia whenever feasible. Libraries should communicate the implications of collections reductions and the importance of reference services to university leadership and researchers.

Clarify the library’s role in supporting the use of generative AI in research and discovery. Libraries should seek opportunities to be campus leaders on issues related to AI and scholarly communication by assessing AI-enabled discovery tools, negotiating licensing and data-use agreements, and making recommendations about the use of AI in scholarly writing.

Make existing research support services more visible to scholars. Libraries should work with research offices and academic units to embed clear referral pathways to reference, discovery, and data services in grant development and other research workflows.

Research Offices

Prepare for future disruptions in federal funding. Research offices should develop transparent criteria for allocating bridge funding, internal grants, and shared resources, with particular consideration for projects, facilities, research teams, and researchers most susceptible to disruption.

Make collaboration a strategic priority and promote broad participation in research. Research offices should simplify agreements and compliance procedures, facilitate the identification of collaborators, and invest in shared infrastructure. Considerations should be made to lower barriers to collaboration for students, research staff, librarians, and other contributors, with clearly defined expectations for roles, compensation, attribution, and mentorship.

Recognize the library as a central partner in research planning and governance. Research offices should collaborate with library leadership and involve the library in decisions related to research strategy, scholarly communication, research support services, and responses to funding disruptions.

Coordinate institution-wide governance of generative AI in research. Research offices should lead the formulation of policies for AI use throughout the research life cycle. This process should involve faculty, librarians, information technology staff, legal counsel, and research integrity personnel.

Funders

Ensure greater continuity in research funding. Funders should provide researchers and institutions with sufficient notice of changes, honor existing commitments, and establish bridge or continuation mechanisms in the event of funding disruptions.

Support the infrastructure necessary for effective collaboration. Private foundations should be prepared to revisit indirect cost policies in response to potential changes in federal policy.

Researchers

Maintain accountability for research conducted with generative AI. Researchers should disclose the use of generative AI, verify the accuracy of AI-generated outputs and citations, and take full responsibility for the integrity of their work.

Set equitable expectations for collaboration and credit. Research teams should establish clear agreements regarding roles, decision-making, authorship, data stewardship, and resource commitments. Contributions from students, staff, librarians, and other collaborators should receive appropriate attribution and compensation.

Acknowledgments

We wish to acknowledge the advisory committee, who helped us identify the core topics covered in the survey and reviewed the survey instrument. All opinions expressed in this report are solely those of Ithaka S+R, and any errors are our own.

• Nicole Branch, Dean of the University Library, Santa Clara University
• Dan Cohen, Vice Provost for Information Collaboration, Dean of the Libraries, and Professor of History, Northeastern University
• Colleen Fitzgerald, Senior Vice Provost for Research and Creative Activities, Hofstra University
• Carlos Rodriguez, Dean of the University Library, California State University, Los Angeles
• Xuemao Wang, Dean of Libraries, Charles Deering McCormick University Librarian, Northwestern University

Appendix: Methodology

Download the survey instrument: https://sr.ithaka.org/wp-content/uploads/2026/09/instrument-9.23.26.docx

Following survey administration, a new variable, Department Category, was created that recoded Discipline Keycode so that respondents with the same Department Code are in the same Department Category. Twenty-seven percent of respondents were recoded. The 20 Department Categories include the following Department Codes, among others: Administrators (Residence Life, Institutional Research, Career Services, Admissions, Registrar, Webmaster, International Services, Athletics Director, Wellness Center Director, Dean, Provost, President); Agriculture (Animal Science, Fisheries, Food Science, Forestry, Horticulture, Plant Pathology, Soil Science, Zoology); Allied Health (Pharmacy, Occupational Therapy, Health Policy, Physical Therapy, Public Health); Business (Accounting, Consumer Science, Finance, Hospitality, Management, Marketing); Communications (Communication Disorders, Journalism, Media Studies); Computer Science; Education (Counseling, Curriculum, Educational Leadership, Human Development, Family Science, Special Education); Engineering (Aeronautics, Biomedical Engineering, Civil Engineering, Electrical Engineering, Industrial Technology, Materials Science, Mechanical Engineering); English Literature (Writing, Rhetoric); Fine Art (Art History, Design, Dance, Music, Film Studies, Performing Arts, Landscape Architecture, Theatre, Illustration); World Language (Classical Studies, East Asian Languages, French, German, Italian, Linguistics, Spanish); General Studies (Arts & Humanities; Arts & Sciences; Honors, Liberal Arts, Professional Studies); Interdisciplinary Studies (Africana Studies, American Studies, Asian Studies, Comparative Literature, Ethnic Studies, Latin American Studies, Women’s Studies); Library Science; Mathematics (Applied Mathematics, Statistics); Medicine (Anatomy, Anesthesiology, Cell Biology, Neurology, Pathology, Pediatrics, Pharmacology, Psychiatry, Surgery); Natural Science (Astronomy, Biology, Chemistry, Earth Science, Environmental Studies, Geology, Marine Science, Physics); Nursing; Physical Education (Exercise Science, Health, Human Performance, Kinesiology, Recreation); Social Science (Anthropology, Criminal Justice, Economics, Geography, History, Law, Military Science, Philosophy, Political Science, Psychology, Religion, Social Work, Sociology).

Endnotes

1. Carly Anne York, “The Rise (and Fall?) of the National Science Foundation,” The Chronicle of Higher Education, February 21, 2025, https://www.chronicle.com/article/the-rise-and-fall-of-the-national-science-foundation. ↑
2. Dylan Ruediger, Chelsea McCracken, and Jonathan Barefield, “The Impact of State and Federal Policies on Academic Researchers: Findings from a National Survey,” Ithaka S+R, April 20, 2026, https://doi.org/10.18665/sr.325485. ↑
3. Respondents who completed 30 percent or more of the survey were included. ↑
4. These classifications are based on the 2021 Carnegie classifications, which were still in effect when the survey was designed and the sample created. ↑
5. See the appendix for more details about how disciplines were coded and for the survey instrument. ↑
6. Reese A. K. Richardson et al., “The Entities Enabling Scientific Fraud at Scale Are Large, Resilient, and Growing Rapidly,” Proceedings of the National Academy of Sciences of the United States of America 122, no. 32 (2025): e2420092122, https://doi.org/10.1073/pnas.2420092122; Kathryn Palmer, “The Growing Problem of Scientific Research Fraud,” Inside Higher Ed, August 12, 2025, https://www.insidehighered.com/news/faculty/research/2025/08/12/growing-problem-scientific-research-fraud; Ritu Phogat et al., “Misconduct in Biomedical Research: A Meta-Analysis and Systematic Review,” Journal of International Society of Preventive & Community Dentistry 13, no. 3 (2023): 185-193, https://www.doi.org/10.4103/jispcd.JISPCD_220_22 ; Jop de Vrieze, “Landmark Research Integrity Survey Finds Questionable Practices Are Surprisingly Common,” Science, July 7, 2021, https://www.science.org/content/article/landmark-research-integrity-survey-finds-questionable-practices-are-surprisingly-common; Sneha Khedkar, “Over 90 Percent of Scientists Admit to Questionable Research Behaviors,” The Scientist, March 25, 2026, https://www.the-scientist.com/over-90-percent-of-scientists-admit-to-questionable-research-behaviors-74258; Maxim Topaz et al., “Fabricated Citations: An Audit across 2·5 Million Biomedical Papers,” The Lancet 407, no. 10541 (May 9, 2026): 1779–81, https://doi.org/10.1016/S0140-6736(26)00603-3. ↑
7. Science and Engineering Indicators, “Funding Sources of Academic R&D,” US National Science Foundation, National Center for Science and Engineering Statistics, October 5, 2023, https://ncses.nsf.gov/pubs/nsb202326/funding-sources-of-academic-r-d. ↑
8. Some of the funding attributed to internal sources was likely at least indirectly federal in origin. This is particularly likely in reference to individuals who received teaching/service release from their institutions, which may have been subsidized by federal grants. ↑
9. Kritika Agarwal, “White House Once Again Proposes Massive Cuts to Scientific Research and Education,” Association of American Universities, April 3, 2026, https://www.aau.edu/newsroom/white-house-once-again-proposes-massive-cuts. ↑
10. Ellen Carroll, Tracy Bergstrom, and Ioana G. Hulbert, “US Library Survey 2025: Under Pressure,” Ithaka S+R, May 14, 2026, https://doi.org/10.18665/sr.325599. ↑
11. Ruby MacDougall and Dylan Ruediger, “The Research Data Services Landscape at US and Canadian Higher Education Institutions,” Ithaka S+R, March 14, 2024, https://doi.org/10.18665/sr.320420. ↑
12. Yusuff Adebayo Adebisi, “Undergraduate Students’ Involvement in Research: Values, Benefits, Barriers and Recommendations,” Annals of Medicine and Surgery 81 (2022): 104384, https://doi.org/10.1016/j.amsu.2022.104384; James Gentile et al. (Eds.), “Undergraduate Research Experiences for STEM Students: Successes, Challenges, and Opportunities,” Washington, DC: The National Academies Press, https://www.nationalacademies.org/read/24622/chapter/5 . ↑
13. Matt Honoré et al., “Contributions Made by Undergraduates to Research Projects: Using the CREDIT Taxonomy to Assess Undergraduate Research Experiences,” Scholarship and Practice of Undergraduate Research 4, no. 1 (2020): 41-51, https://doi.org/10.18833/spur/4/1/3. ↑
14. M. A. Gouda et al., “Medical Undergraduates’ Contributions to Publication Output of World’s Top Universities in 2013,” QJM: An International Journal of Medicine 109, no. 9 (2016): 605–11, https://doi.org/10.1093/qjmed/hcw028; Traci A. Giuliano, “How Common Is Undergraduate Publication in Psychology? An Examination of Faculty Vitae from Top Colleges and Universities,” Collabra: Psychology 9, no. 1 (2023): 84521, https://doi.org/10.1525/collabra.84521. ↑
15. Jeffrey Mervis, “NSF Downsizes Summer Research Program for Undergraduates,” Science, February 27, 2025, https://www.science.org/content/article/nsf-downsizes-summer-research-program-undergraduates; Max Kozlov et al., “Trump Administration Proposes Massive Budget Cuts to Science,” Scientific American, April 4, 2026, https://www.scientificamerican.com/article/trump-administration-proposes-massive-budget-cuts-to-science/. ↑
16. Dylan Ruediger et al., “Big Data Infrastructure at the Crossroads: Support Needs and Challenges for Universities,” Ithaka S+R, December 1, 2021, https://doi.org/10.18665/sr.316121. ↑
17. Dylan Ruediger, “Federal Funding Cuts and Research Universities: New Tracker Documents the Impacts of the Cuts on the Human Infrastructure of the Research Enterprise,” Ithaka S+R, October 22, 2025, https://sr.ithaka.org/blog/federal-funding-cuts-and-research-universities/; “University Research Workforce Tracker,” Ithaka S+R, https://sr.ithaka.org/our-work/the-research-enterprise/university-research-workforce-tracker/. ↑