Ma, Ji 马季

Office hour: Tue 2pm–4pm (2025 spring).
Office: SRH3.324, +1-512-232-4240
Email: [email protected] (preferred) | [email protected]

My name is Ji MA (马季; first name pronounces “G”), Assistant Professor at the Lyndon B. Johnson School of Public Affairs at the University of Texas at Austin; Visiting Researcher at the Gradel Institute of Charity, University of Oxford. I’m also an affiliated faculty member of the School of Information, the Center for East Asian Studies, and the Asia Policy Program at UT Austin.

I focus on the nexus of technology and society through three interrelated research streams: (1) Advanced Technologies for the public and nonprofit sectors and Computational Methods for social science research; (2) Knowledge Production for evidence-based policymaking and nonprofit studies; (3) State-Society Relations in China. My publications appear in leading journals across a broad range of disciplines, including public administration and policy, nonprofit studies, sociology, political science, and data science.

In my spare time, I enjoy building open-source projects, tasting tea and hiking with family and friends, and practicing Olympic Recurve.

selected publications

2025

Steering Prosocial AI Agents: Computational Basis of LLM’s Decision Making in Social Simulation

Ji Ma

Apr 2025

Abs DOI

Large language models (LLMs) increasingly serve as human-like decision-making agents in social science and applied settings. These LLM-agents are typically assigned human-like characters and placed in real-life contexts. However, how these characters and contexts shape an LLM’s behavior remains underexplored. This study proposes and tests methods for probing, quantifying, and modifying an LLM’s internal representations in a Dictator Game – a classic behavioral experiment on fairness and prosocial behavior. We extract “vectors of variable variations” (e.g., “male” to “female”) from the LLM’s internal state. Manipulating these vectors during the model’s inference can substantially alter how those variables relate to the model’s decision-making. This approach offers a principled way to study and regulate how social concepts can be encoded and engineered within transformer-based models, with implications for alignment, debiasing, and designing AI agents for social simulations in both academic and commercial applications.
Why do some academic articles receive more citations from policy communities?

Ji Ma and Yuan (Daniel) Cheng

Public Administration Review, Apr 2025

Abs DOI

We (1) present the landscape of the citations of Public Administration and Policy (PAP) scholarly articles in policy documents and (2) examine influencing factors along three dimensions: collaborative teams, cross-disciplinary interactions, and disruptive paradigms. Using data from the 30 most-cited PAP peer-reviewed journals and 38,062 documents from 1107 policy institutions, we find that 10.1% of all PAP scholarship receives high citations from both academics and policy communities. Collaborative teams, cross-disciplinary interactions, and disruptive paradigms can all increase the citations within policy communities, yet the relationships are not linear. Nonacademic authors can consistently attract more policy citations, whether publishing alone or collaborating with academics. An article should ideally cite no more than 13 disciplinary subjects. No significant trade-off between scholarly and policy impact as scholarly citations and the academic reputation of authors often translate into policy citations. These findings offer novel and concrete insights into optimizing academic research for policy impact.

2024

Consensus Formation in Nonprofit and Philanthropic Studies: Networks, Reputation, and Gender

Ji Ma and René Bekkers

Nonprofit and Voluntary Sector Quarterly, Feb 2024

Abs DOI

The research field of nonprofits and philanthropy has grown exponentially. To what extent do nonprofit scholars share a common language? Answering this question is crucial to assessing the field’s intellectual cohesiveness. We studied how coauthor networks, scholarly reputation, and the prevalence of female authors influence consensus formation. We found that the degree of consensus for all major research topics in the field has increased over time—For every 10% growth in the volume of literature, shared language increased by 1.4%. A cohesive research community on nonprofits and philanthropy has been forming since the early 2000s. Female scholars are fewer in number and less cited than males; their presence did not exceed 40% for most topics. The citation counts of scholars and small-world property of networks are positively associated with consensus, suggesting that star researchers and knowledge brokers bridging different intellectual communities are key to sharing research interests and language.
Neutral, Non-Disruptive, and Native: Why Do Chinese Nonprofit Scholars Cite English Articles?

Ji Ma

Nonprofit and Voluntary Sector Quarterly, Aug 2024

Abs DOI

Language shapes diverse cultures and creates natural barriers between human societies. The landscape of nonprofit and philanthropic studies in non-English languages is barely charted, impeding the globalization of this research field. This project (a) describes the topics shared between English and Chinese scholarship on nonprofits and philanthropy and (b) explores why English scholarship is cited in Chinese journal articles from five aspects: rationale of scholarship, novelty, relevance, social network, and reputation. The English articles cited by Chinese scholars tend to: (a) focus on instrumentality but not expressive values, (b) develop rather than disrupt existing paradigms, and (c) be relevant to topics popular in Chinese literature and have authors with Chinese scholarly connections. In general, Chinese scholars tend to cite English articles that are value-neutral, non-disruptive, and native. Theoretical and methodological implications for examining nonprofit studies in other languages are discussed.
Can Machines Think Like Humans? A Behavioral Evaluation of LLM-Agents in Dictator Games

Ji Ma

Dec 2024

Abs DOI

As Large Language Model (LLM)-based agents increasingly undertake real-world tasks and engage with human society, how well do we understand their behaviors? We (1) investigate how LLM agents’ prosocial behaviors – a fundamental social norm – can be induced by different personas and benchmarked against human behaviors; and (2) introduce a behavioral and social science approach to evaluate LLM agents’ decision-making. We explored how different personas and experimental framings affect these AI agents’ altruistic behavior in dictator games and compared their behaviors within the same LLM family, across various families, and with human behaviors. The findings reveal substantial variations and inconsistencies among LLMs and notable differences compared to human behaviors. Merely assigning a human-like identity to LLMs does not produce human-like behaviors. Despite being trained on extensive human-generated data, these AI agents are unable to capture the internal processes of human decision-making. Their alignment with human is highly variable and dependent on specific model architectures and prompt formulations; even worse, such dependence does not follow a clear pattern. LLMs can be useful task-specific tools but are not yet intelligent human-like agents.

2023

Computational Social Science for Nonprofit Studies: Developing a Toolbox and Knowledge Base for the Field

Ji Ma, Islam Akef Ebeid, Arjen Wit, and 4 more authors

VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, Feb 2023

Abs DOI

How can computational social science (CSS) methods be applied in nonprofit and philanthropic studies? This paper summarizes and explains a range of relevant CSS methods from a research design perspective and highlights key applications in our field. We define CSS as a set of computationally intensive empirical methods for data management, concept representation, data analysis, and visualization. What makes the computational methods “social” is that the purpose of using these methods is to serve quantitative, qualitative, and mixed-methods social science research, such that theorization can have a solid ground. We illustrate the promise of CSS in our field by using it to construct the largest and most comprehensive database of scholarly references in our field, the Knowledge Infrastructure of Nonprofit and Philanthropic Studies (KINPS). Furthermore, we show that through the application of CSS in constructing and analyzing KINPS, we can better understand and facilitate the intellectual growth of our field. We conclude the article with cautions for using CSS and suggestions for future studies implementing CSS and KINPS.
How Does an Authoritarian State Co-opt Its Social Scientists Studying Civil Society?

Ji Ma

VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, Aug 2023

Abs DOI

What channels can an authoritarian state employ to steer social science research towards topics preferred by the regime? I researched the Chinese coauthor network of civil society studies, examining 14,088 researchers and their peer-reviewed journal articles published between 1998 and 2018. Models with individual and time fixed-effects reveal that scholars at the center of the network closely follow the narratives of the state’s policy plans and could serve as effective state agents. However, those academics who connect different intellectual communities tend to pursue novel ideas deviating from the official narratives. Funding is an ineffective direct means for co-opting individual scholars, possibly because it is routed through institutions. Combining these findings, this study proposes a preliminary formation of authoritarian knowledge regime that consists of (1) the state’s official narrative, (2) institutionalized state sponsorship, (3) co-opted intellectuals centrally embedded in scholarly networks, and (4) intellectual brokers as sources of novel ideas.

2021

Automated Coding Using Machine Learning and Remapping the U.S. Nonprofit Sector: A Guide and Benchmark

Ji Ma

Nonprofit and Voluntary Sector Quarterly, Jun 2021

Abs DOI

This research developed a machine learning classifier that reliably automates the coding process using the National Taxonomy of Exempt Entities as a schema and remapped the U.S. nonprofit sector. I achieved 90% overall accuracy for classifying the nonprofits into nine broad categories and 88% for classifying them into 25 major groups. The intercoder reliabilities between algorithms and human coders measured by kappa statistics are in the “almost perfect” range of .80 to 1.00. The results suggest that a state-of-the-art machine learning algorithm can approximate human coders and substantially improve researchers’ productivity. I also reassigned multiple category codes to more than 439,000 nonprofits and discovered a considerable amount of organizational activities that were previously ignored. The classifier is an essential methodological prerequisite for large-N and Big Data analyses, and the remapped U.S. nonprofit sector can serve as an important instrument for asking or reexamining fundamental questions of nonprofit studies. The working directory with all data sets, source codes, and historical versions are available on GitHub (https://github.com/ma-ji/npo_classifier).

2018

A Century of Nonprofit Studies: Scaling the Knowledge of the Field

Ji Ma and Sara Konrath

VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, Dec 2018

Abs DOI

This empirical study examines knowledge production between 1925 and 2015 in nonprofit and philanthropic studies from quantitative and thematic perspectives. Quantitative results suggest that scholars in this field have been actively generating a considerable amount of literature and a solid intellectual base for developing this field toward a new discipline. Thematic analyses suggest that knowledge production in this field is also growing in cohesion—several main themes have been formed and actively advanced since 1980s, and the study of volunteering can be identified as a unique core theme of this field. The lack of geographic and cultural diversity is a critical challenge for advancing nonprofit studies. New paradigms are needed for developing this research field and mitigating the tension between academia and practice. Methodological and pedagogical implications, limitations, and future studies are discussed.
State power and elite autonomy in a networked civil society: The board interlocking of Chinese non-profits

Ji Ma and Simon DeDeo

Social Networks, Jul 2018

Abs DOI

In response to failures of central planning, the Chinese government has experimented not only with free-market trade zones, but with allowing non-profit foundations to operate in a decentralized fashion. A network study shows how these foundations have connected together by sharing board members, in a structural parallel to what is seen in corporations in the United States and Europe. This board interlocking leads to the emergence of an elite group with privileged network positions. While the presence of government officials on non-profit boards is widespread, government officials are much less common in a subgroup of foundations that control just over half of all revenue in the network. This subgroup, associated with business elites, not only enjoys higher levels of within-elite links, but even preferentially excludes government officials from the NGOs with higher degree. The emergence of this structurally autonomous sphere is associated with major political and social events in the state–society relationship. Cluster analysis reveals multiple internal components within this sphere that share similar levels of network influence. Rather than a core-periphery structure centered around government officials, the Chinese non-profit world appears to be a multipolar one of distinct elite groups, many of which achieve high levels of independence from direct government control.

2017

The research infrastructure of Chinese foundations, a database for Chinese civil society studies

Ji Ma, Qun Wang, Chao Dong, and 1 more author

Scientific Data, Jul 2017

Abs DOI

This paper provides technical details and user guidance on the Research Infrastructure of Chinese Foundations (RICF), a database of Chinese foundations, civil society, and social development in general. The structure of the RICF is deliberately designed and normalized according to the Three Normal Forms. The database schema consists of three major themes: foundations’ basic organizational profile (i.e., basic profile, board member, supervisor, staff, and related party tables), program information (i.e., program information, major program, program relationship, and major recipient tables), and financial information (i.e., financial position, financial activities, cash flow, activity overview, and large donation tables). The RICF’s data quality can be measured by four criteria: data source reputation and credibility, completeness, accuracy, and timeliness. Data records are properly versioned, allowing verification and replication for research purposes.