ISOM Faculty Research

Costello Research Business Model Innovation

Research

Can machine learning make the world a fairer place?

Katelynn C Hipolito — Wed, 22 Apr 2026 15:08:00 +0000

Can machine learning make the world a fairer place? Katelynn C Hipolito Wed, 04/22/2026 - 11:08

A paradox hovers over our increasingly AI-dependent world. On the one hand, artificial intelligence can make the world a better place (or so we’re told). On the other hand, algorithms have no imagination or consciousness, and thus can know only the status quo—as reflected in the data they are trained on. And our current world is far from perfectly meritocratic or fair.

Jingyuan Yang. Photo by Jeffrey Porovich/Costello College of Business.

Jingyuan Yang, assistant professor of information systems and operations management at Costello College of Business at 91��, suggests that the paradox is compounded by conventional thinking around AI. “The standard view is that fairness is a tax on efficiency. The way conventional systems are structured, fairness checks are added almost as an afterthought that is assumed to negatively impact system performance,” she says.

Is the “better,” optimized world of AI destined to replicate, or perhaps even exacerbate, existing inequalities? Yang’s ongoing research—in collaboration with Pengzhan Guo of Duke Kunshan University and Keli Xiao of Stony Brook University—points to an appealing alternative. It uses AI systems as a proving ground for a theorized “fairness-performance complementarity”—the idea that, under certain conditions, fairness and performance reinforce one another.

“Our 'fairness-by-design’ framework utilizes reinforcement learning, which is a type of machine learning (ML). But unlike most machine learning algorithms, ours includes multiple agents competing for finite resources in a dynamic environment, not a static one,” Yang says. “That makes our paradigm much more structurally similar to many real-world environments in which various people compete over time for finite resources.”

Fairness was integrated in two stages. First, the framework was designed to “nudge” high-performing agents towards exploratory choices that might maximize their rewards. As Yang explains, “In this framework, high-performing agents are held in an exploratory mode for longer, while lower-performing agents settle into stable paths sooner.” Second, options that were abandoned as a result of agents’ reward-seeking behavior were redistributed, with lower-performing agents getting first crack at the best opportunities.

As Yang summarizes, "The exploratory activity of the high performers releases opportunities that the system channels down toward the weaker performers. Theoretically, this increases fairness while retaining individual choice and without constraining performance.”

“Our ‘fairness-by-design’ framework utilizes reinforcement learning, which is a type of machine learning (ML). But unlike most machine learning algorithms, ours includes multiple agents competing for finite resources in a dynamic environment, not a static one. That makes our paradigm much more structurally similar to many real-world environments in which various people compete over time for finite resources.”

—Jingyuan Yang, assistant professor of information systems and operations management at Costello College of Business at 91��

To test out the framework, the researchers used a data-set comprising detailed information on the job histories of 6.5 million professionals across a 20-year timeframe. “In the real-world data, we see a high degree of disparity, without very much redistribution of elite opportunities from relatively advantaged to disadvantaged employees,” Yang says.

The algorithm converted the real-world job information into opportunities offered to hypothetical agents. The resulting career paths were analyzed in terms of both performance and fairness. Performance was defined by aggregate rewards earned by all agents across all periods. Fairness was defined by the degree to which initial performance disparities were resolved over successive decisions.

The “fairness-by-design” framework’s results—for both fairness and performance—were better than those of eight alternative ML methods drawn from three different methodological families.

The researchers also adjusted the system to account for people’s changing preferences. Early-career professionals tend to value employer reputation and advancement potential; in late career, rewards pertaining to job stability and security are more salient. Even with these restrictions implemented, the framework functioned as intended—improving the average quality of overall career paths while fueling upward mobility.

In a follow-up study utilizing the New York Yellow Taxi Trip record database, the framework was tasked with generating route recommendations to hypothetical “agents,” i.e. cab drivers, with varying performance records. In this domain, the choice-set was much smaller (263 locations, as compared to 4,282 companies), and the timeframe far shorter (two hours as opposed to 20 years). As with the career-planning example, the taxi study found that more equitable distribution of high-quality routes led to higher average income per minute for the system as a whole.

“Because the framework proved adaptable to different domains and agent preferences, we think it could be used in future as a governance mechanism for a variety of AI contexts,” Yang says. Health care scheduling, course registration in higher education and provision of digital services are a few areas Yang sees as likely candidates.

While emphasizing that her research is still ongoing, she argues that it poses a serious challenge to standard ways of thinking about AI. “Our formal proof establishes the conditions under which fairness and performance reinforce each other, and our experiments show those conditions are achievable in realistic settings. That gives our work both theoretical and experimental grounding," Yang says.

Learn more about AI at George Mason

Dive into Research at George Mason

In This Story

Jingyuan Yang

Topics

Costello Research Artificial Intelligence

Costello Research Machine Learning

Online ad fraud is a feature, not a bug

Katelynn C Hipolito — Tue, 17 Mar 2026 17:26:46 +0000

Online ad fraud is a feature, not a bug Katelynn C Hipolito Tue, 03/17/2026 - 13:26

With print media circulation and broadcast television viewership in free fall, a lot is riding on the online advertising space being able to take up the slack. The good news is, digital ad spend is booming: The global total for 2025 is expected to surpass $678 billion, at an annual growth rate of nearly eight percent.

The bad news? A good chunk of that money is chasing a mirage.

Online ad fraud—where ad publishers falsely inflate engagement metrics (impressions, clicks, etc.) to boost revenues—is a growing problem that eats upwards of 20 percent of global ad spend.

(Left to Right) Abhishek Ray and Min Chen. Photo by Hannah Patterson/Costello College of Business.

Min Chen and Abhishek Ray, both professors in the information systems and operations management area at Costello College of Business at 91��, are researching how online ad networks, such as Google Ads, can improve upon existing anti-fraud methods. Their recently published paper in Management Science explores deep-rooted dynamics of the online ad ecosystem that make eliminating fraud even more complicated than it may seem at first glance. The paper was co-authored by Subodha Kumar of Temple University.

The researchers used a game-theoretic model to replicate the interconnected decision-making of the three players involved: advertisers, publishers, and the networks that serve as go-between.

“The way the ecosystem works is that the platforms in the middle, the ad networks, shares the benefit from the transaction,” Chen explains. “People have been arguing whether the network is incentivized to put their best efforts behind deterring fraud, since the fraudulent traffic benefits the networks too. So we tried to create a model to capture this.”

“If the advertisers rely solely on the reports from the ad networks, they may be at risk. They should use third-party tools to audit the performance better.”

—Min Chen, information systems and operations management professor at the Costello College of Business at 91��

In addition, the model incorporates the two main fraud deterrents that networks routinely use. One is technological—platforms can adopt tougher standards for fraud detection, widening the scope of suspicious activity that gets flagged. The other is economic—lowering payments to all publishers so as to disincentivize large-scale fraud.

Surprisingly, the researchers find that the online ad economy works best when the two approaches seem to be working at cross-purposes. A tightening in fraud detection technology, paired with high payments for publishers, may sometimes produce the best outcomes for advertisers, publishers, and networks, as the market evolves.

The reason is rooted in the imperfect nature of fraud detection. To be sure, detection systems are improving all the time, especially with the advent of AI. But fraudsters do their best to blend in and adapt, using technological tools that often outpace those of their pursuers. “You cannot catch all the fraud, and if you try, you are going to mis-detect a lot of non-fraud,” Chen says.

Tougher fraud detection, then, will always mean more false positives, no matter how good the technology gets. To counter this inherent unfairness that penalizes good and bad actors alike, the ad network’s payment to publishers need to go up. Otherwise, publishers may take their business elsewhere—especially those most valuable to the system, i.e. those that are trustworthy— thereby decreasing the advertisers’ valuation on ad traffic.

“These ad networks are kind of a unique system where you can be monetarily rewarded for being honest, or punished for being dishonest,” Ray says. “What we discover for this system is there can be a way in which we can give carrots to people, not just sticks.”

On a similar note, the researchers find that an attempt to purge “bad apple” advertisers from the system can backfire due to false positives. In fact, fraud can sharply increase if networks, believing they have solved the problem, relax their fraud detection standards and raise incentives for the remaining advertisers. “Since the publishers who produce the fraudulent traffic are fewer now, the ad network may no longer need to maintain a strict detection policy. This can encourage the remaining ones to commit much more fraud,” Chen explains.

To Ray and Chen, online ad fraud is, in at least one sense, no different from older forms of malfeasance that are found in all free societies. “We need to have some kind of mechanism for managing the level of fraud, because the fraud detection method is never going to be perfect, whether it’s financial fraud, accounting fraud, etc.,” Chen says.

But as an example of the contemporary platform economy, the online advertising ecosystem is also distinctive, in that its de facto regulatory authority has skin in the game. The ad networks’ mixed incentives—as both beneficiaries and inhibitors of fraud—can undermine integrity and trust within an already-compromised system.

“If the advertisers rely solely on the reports from the ad networks, they may be at risk,” Chen says. “They should use third-party tools to audit the performance better.”

Learn more about the Costello College of Business

Discover Research at George Mason

In This Story

Min Chen

Abhishek Ray

Topics

Costello Research Data Analytics

Are salespeople more effective when they’re being monitored?

Katelynn C Hipolito — Tue, 16 Dec 2025 15:28:57 +0000

Are salespeople more effective when they’re being monitored? Katelynn C Hipolito Tue, 12/16/2025 - 10:28

How employees respond to being under surveillance depends on a number of factors, including how good they are at their jobs.

Digital and online technologies have made our workplace routines faster and easier. They have also made it easier for managers to keep tabs on workers, via monitoring apps designed to capture whether employees are “working hard, or hardly working.”

Brad Greenwood. Photo provided.

But for researchers such as Brad Greenwood and Ioannis Bellos, professors in the information systems and operations management (ISOM) area at 91��’s Costello College of Business, the jury’s still out on whether the latest worker surveillance tech actually benefits performance, productivity, efficiency, etc., especially in certain key sectors.

“Prior research has identified both harms and benefits stemming from worker surveillance,” they state. “But very little research has been done in the retail or sales space, especially in a real-life sales environment as opposed to the lab.”

In contrast, Bellos and Greenwood’s forthcoming academic paper in Manufacturing and Service Operations Management homes in on one China-based company, an online supplier of baby products (diapers, formula, etc.) to physical retail stores. This B2B company employs hundreds of salespeople to partner directly with small-business owners, not only on promotion and upselling but also helping solve marketing and operational challenges related to the product line. Naturally, the sales force’s job description includes a lot of in-person, on-site collaboration.

The paper was co-authored by Yingda Lu of University of Illinois at Chicago and Liqiang Huang of Zhejiang University.

In 2019, the company issued phones equipped with GPS technology to track the frequency and length of salespeople’s site visits. This was not a secret program; sales teams were told they would be tracked via GPS. Due to a technical glitch, not all salespeople received the phones at the same time. The erratic rollout allowed the research team to draw clean comparisons between monitored and unmonitored employees within the same timeframe.

Almost immediately, monitored salespeople began making more client visits, spending longer at each store and engaging in a wider variety of tasks each time. As a result, overall sales performance—i.e., the gross merchandise value (GMV) of goods sold to clients—rose about 4.75 percent over the pre-surveillance average.

Ioannis Bellos. Photo by Office of University Branding

But that wasn’t the whole story. The GMV for top-performing salespeople went down once they knew they were being monitored, despite the fact that they—like the rest of the sales force—ramped up their sales activity to look good for the GPS. The critical difference was that top performers had successful routines before the GPS came in. Once they started performing for Big Brother, rather than continuing with their familiar working patterns, their outcomes started to suffer.

“If you tell underperformers ‘We’re going to monitor you,’ it lights a fire under them,” Greenwood says. “But if you tell high performers the same thing, they try to guess what they think you want them to do and do that rather than just doing their job.”

The researchers discovered that engaging in more sales calls resulted in diminishing returns for the best salespeople, because they were spreading their skills too thin across a less profitable pool of clients.

Bellos and Greenwood envision that declining commissions and a disrupted work routine for top performers could lead to negative outcomes for the company. For example, some star salespeople could become disengaged and even start looking for work elsewhere.

“When it comes to these sorts of MVPs, anything that harms their relationship with the firm opens the door to some very real negatives,” Greenwood says.

For Greenwood and Bellos, this speaks to the need for companies to be nuanced and strategic in their approaches to worker surveillance. “Whether monitoring works or doesn’t work will have a lot to do with individuals,” they state. “When we treat people like monoliths, it’s hard to understand what’s really going on. You can improve high-level performance while at the same time sub-optimizing other aspects of the firm.”

Sometimes, proper communication and clarity can make monitoring less problematic, the researchers contend. If employees are fully briefed on the goals and uses of monitoring programs, they won’t have to rely on guesswork as they adjust their routines. This is especially true for workers with a diverse and complicated portfolio of tasks, such as the sales force in Bellos and Greenwood’s study.

“In our study, salespeople had a high degree of freedom in that within each store, they have several tasks to perform. Plus, each salesperson services multiple stores,” says Bellos. “Managers and platforms need to factor [task diversity] in on the implementation side.”

“In our study, salespeople had a high degree of freedom in that within each store, they have several tasks to perform. Plus, each salesperson services multiple stores. Managers and platforms need to factor [task diversity] in on the implementation side.”

— Ioannis Bellos, Director of Master of Business Administration and Master of Science in Management Programs, and Business Certificates

In This Story

Brad Greenwood

Ioannis Bellos

Topics

Costello Research Innovation Strategy

George Mason researcher is using AI to identify human trafficking hot spots

Katelynn C Hipolito — Mon, 29 Sep 2025 18:20:46 +0000

George Mason researcher is using AI to identify human trafficking hot spots Katelynn C Hipolito Mon, 09/29/2025 - 14:20

Illicit massage businesses (IMBs) run by human trafficking rings are rampant in the United States. A 91�� professor has helped build what may be the best artificial intelligence (AI)-driven tool to root them out.

Human trafficking rings are at their most dangerous when they masquerade as legitimate commercial activity. IMBs are one of the most common ways in which exploitive networks operate in plain sight.

The Network, an anti-trafficking nonprofit, estimates that there are more than 13,000 IMBs active in the United States, raking in annual total revenue of more than $5 billion.

Abhishek Ray

“You are stuck in a massage business. You’re not allowed to go out,” says Abhishek Ray, assistant professor of information systems and operations management at the Donald G. Costello College of Business at George Mason, describing the plight of IMB workers. “Your passports are taken away, and you’re supposed to do a certain amount of business every day and give the money to the trafficker. It’s a really abhorrent form of abuse.”

Ray is one of a growing group of researchers exploring how various forms of AI could help resource-constrained law enforcement agencies differentiate between IMBs and the legitimate enterprises they try to mimic. His ongoing research using graph neural networks has yielded more promising results than rival approaches, when put to the test in a recent experiment.

His co-researchers on the IMB project are Lumina Albert and Swetha Varadarajan of Colorado State University.

According to Ray, “Graph neural networks are just a fancy way of saying that if I get a graph of a city or locality at one point in time, and I add data to it, can I predict future patterns on this graph if I know the past?”

This approach made sense for detecting IMBs, because try as they might to appear above board, they have geographical needs that conventional businesses don’t. “IMBs don’t allow their trafficked employees to go out of the parlor,” Ray says. “But since they’re humans, they need sustenance. They have to be near groceries, gas stations, where they can get stuff and come back.”

The researchers combined several graph neural networks into a framework called IMBWatch. The training data-set for IMBWatch comprised publicly available information such as online customer reviews, arrest and raid data for known IMBs, and advertisements from websites promoting illicit activities (e.g., the infamous Backpage). The result, in essence, was a series of snapshots mapping the evolution of the IMB network in a given city or county over a period of time. This could then be overlaid on geographical maps to tease out hidden patterns.

To gauge IMBWatch’s relative performance, the researchers let it loose on a testing data-set alongside four other AI models, which were not as sensitive to the nuanced interplay of spatial and temporal factors. Of the five models, IMBWatch provided the most accurate, precise and informative predictions. In other words, it outperformed the others at spotting IMBs among a larger mass of local businesses.

“Graph neural networks are just a fancy way of saying that if I get a graph of a city or locality at one point in time, and I add data to it, can I predict future patterns on this graph if I know the past?”
— Abhishek Ray, assistant professor of information systems and operations management

While encouraging, these outcomes require further confirmation with a larger data-set. “IMBWatch was trained on data from Georgia and Louisiana, not the entire United States,” Ray says. “These were small, manageable data-sets, but we will now scale up to major states such as New York and California.”

The researchers are also looking at enhancing IMBWatch with data related to how workers end up wandering into trafficking webs. These might include “proximity to hospitals, religious places, etc. because a lot of times people are coerced by religious compulsions, or because they’re pregnant and need some care,” Ray says.

This is not Ray’s first foray into the field of AI-fueled anti-trafficking. Previously, Ray co-developed a model for improving machine learning-based detection of human trafficking activity at transit stations and on fishing vessels.

However, law enforcement agencies and other pertinent stakeholders (e.g., business owners) are often wary of adopting AI-based solutions, due to a lack of trust in the technology. Ray and his co-researchers are currently devising a framework that will clarify how these stakeholders can work together with tech experts and, perhaps most importantly, human trafficking survivors to make the best possible use of AI.

“This qualitative piece is required to make sure that people who are on the sidelines, on the fences about using this actually start using it, because that’s the need right now,” Ray says.

Learn more about Artificial Intelligence at George Mason

In This Story

Abhishek Ray

This content appears in the Spring 2026 print edition of the Mason Spirit Magazine with the title "Using AI to Identify Human Trafficking Hot Spots."

Nonprofits are in trouble. Could more sensitive chatbots be the answer?

Jennifer Anzaldi — Tue, 18 Mar 2025 14:48:25 +0000

Nonprofits are in trouble. Could more sensitive chatbots be the answer? Jennifer Anzaldi Tue, 03/18/2025 - 10:48

In today’s attention economy, impact-driven organizations are arguably at a disadvantage. Since they have no tangible product to sell, the core of their appeal is emotional rather than practical—the “warm glow” of contributing to a cause you care about. But emotional appeals call for more delicacy and precision than standardized marketing tools, such as mass email campaigns, can sustain. Emotional states vary from person to person—even from moment to moment within the same person.

Photo by Getty Images

Siddharth Bhattacharya and Pallab Sanyal, professors of information systems and operations management at the Donald G. Costello College of Business at 91��, believe that artificial intelligence (AI) can help solve this problem. A well-designed chatbot could be programmed to calibrate persuasive appeals in real time, delivering messaging more likely to motivate someone to take a desired next step, whether that’s donating money, volunteering time or simply pledging support. Automated solutions, such as chatbots, can be especially rewarding for nonprofits, which tend to be cash-conscious and resource-constrained.

“We completed a project in Minneapolis and are working with other organizations, in Boston, New Jersey and elsewhere, but the focus is always the same,” Sanyal says. “How can we leverage AI to enhance efficiency, reduce costs, and improve service quality in nonprofit organizations?”

Siddarth Bhattacharya. Photo provided

Sanyal and Bhattacharya’s working paper (coauthored by Scott Schanke of University of Wisconsin Milwaukee) describes their recent randomized field experiment with a Minneapolis-based women’s health organization. The researchers designed a custom chatbot to interact with prospective patrons through the organization’s Facebook Messenger app. The bot was programmed to adjust, at random, its responses to be more or less emotional, as well as more or less anthropomorphic (human-like).

“For the anthropomorphic condition, we introduced visual cues such as typing bubbles and slightly delayed response to mimic the experience of messaging with another human,” Sanyal says.

The chatbot’s “emotional” mode featured more subjective, generalizing statements with liberal use of provocative words such as “unfair,” “discrimination” and “unjust.” The “informational” modes leaned more heavily on facts and statistics.

Over the course of hundreds of real Facebook interactions, the moderately emotional chatbot achieved deepest user engagement, as defined by a completed conversation. (Completion rate was critical because after the last interaction, users were redirected to a contact/donation form.) But when the emotional level went from moderate to extreme, more users bailed out on the interaction.

The takeaway may be that “there is a sweet spot where some emotion is important, but beyond that emotions can be bad,” as Bhattacharya explains.

Pallab Sanyal. Photo provided

When human-like features were layered on top of emotionalism, that sweet spot got even smaller. Anthropomorphism lowered completion rates and reduced the organization’s ability to use emotional engagement as a motivational tool.

“In the retail space, studies have shown anthropomorphism to be useful,” Bhattacharya says. “But in a nonprofit context, it’s totally empathy-driven and less transactional. If that is the case, maybe these human cues coming from a bot make people feel creepy, and they back off.”

Sanyal and Bhattacharya say that more customized-chatbot experiments with other nonprofits are in the works. They are taking into careful consideration the success metrics and unique needs of each partner organization.

“Most of the time, we researchers sit in our offices and work on these problems,” Sanyal says. “But one aspect of these projects that I really like is that we are learning so much from talking to these people.”

In collaboration with the organizations concerned, they are designing chatbots that can cater their persuasive appeals more closely to each context and individual interlocutor. If successful, this method would prove that chatbots could become more than a second-best substitute for a salaried human being. They could serve as interactive workshops for crafting and refining an organization’s messaging to a much more granular level than previously possible.

And this would improve the effectiveness of organizational outreach across the board—a consummate example of AI enhancing, rather than displacing, human labor. “This AI is augmenting human functions,” says Sanyal. “It’s not replacing. Sometimes it’s complementing, sometimes it’s supplementing. But at the end of the day, it is just augmenting.”

In This Story

Siddharth Bhattacharya

Pallab Sanyal

Topics

Costello Research Artificial Intelligence

Costello Research Machine Learning

Costello College of Business health care research puts “patients at the center”

Jennifer Anzaldi — Tue, 11 Mar 2025 15:55:18 +0000

Costello College of Business health care research puts “patients at the center” Jennifer Anzaldi Tue, 03/11/2025 - 11:55

In This Story

Nirup Menon

Like virtually every other industry, health care is increasingly prioritizing digital transformation. The sector is unique, however, in that its results are measured not only in business terms but also tangible outcomes for people—often, literal life and death. So are newly acquired technologies actually paying off for patients?

Nirup Menon, a professor of information systems at the Donald G. Costello College of Business at 91��, says that the answer is “not always.”

Nirup Menon

His recently published paper in Decision Support Systems tackles the so-called “HIT paradox,” or the widespread perception that health information technologies (HIT) have not yet moved the needle on important outcomes such as productivity, quality of care, and patient safety.

Menon co-authored the paper with Costello colleagues Amitava Dutta and Sidhartha Das.

Based on comprehensive survey data from approximately 6,000 U.S. hospitals, the research team looked into whether those that adopted Clinical Decision Support Systems (CDSS) saw lower mortality rates for cardiac patients.

“CDSS is not only for cardiologists,” Menon explains. “It is hospital-based—a system that helps with clinical decision-making. But we know that many cardiac patients may not necessarily have cardiac as their only problem. There are probably decisions being made about them using all kinds of ailments and medications, and so on.”

The basic idea behind CDSS is to use technology to mine actionable insights from a wealth of patient data, giving clinicians key tools to make informed decisions at the point of care. Theoretically, a hospital with CDSS solutions should be much better equipped to handle complex cases—such as a heart-attack sufferer with diabetes or another comorbidity—in real time than one without.

However, Menon and his co-authors discovered that when it came to preventing deaths from cardiac emergencies, the impact of CDSS was context-specific. Their paper finds a number of complementary effects suggesting that health care technologies need help from their environment in order to be most effective. For example, the presence of cardiac medical services (CMS), e.g. diagnostic catheterization and electrophysiology, was unsurprisingly associated with lower mortality rates—but CMS combined with CDSS was more impactful than either on its own.

“The labor force—by which I mean the physician and the entire team of nurses and technicians—should be trained to use this technology appropriately,” Menon summarizes. “You also need real-time integration between CDSS and other IT systems, because if it’s not well-integrated, the provider will not have all the data at their fingertips. If you don’t provide the right inputs into a CDSS, it’s not going to give you the right outputs.”

Menon points out that the “HIT paradox” isn’t limited to CDSS or any single technology. President Obama’s 2009 economic stimulus package, after all, included tens of billions in financial incentives for health care providers to digitize their patient records. By 2017, 95 percent of U.S. hospitals had adopted electronic patient records. Yet, as Menon tells it, “hospitals are just chugging along. The quality remains the same and the costs are just increasing. Or you might see improvements in one small department. So we are trying to find the variables that create complementarities within large samples.”

Menon knows, however, that the applications of health care tech can be closely targeted to relatively tiny patient populations, too. Another recent paper of his, published in JMIR Medical Informatics, uses causal survival forests, a machine-learning algorithmic technique, to determine which of two chemotherapy drugs promoted the most longevity for terminal prostate cancer patients. Taking into account age, race and comorbidity symptoms, their analysis produced an easy-to-use prescription policy tree that, by itself, could extend patients’ lives by almost two months—if the test sample, comprised of 2,886 veterans treated at VA health centers, was representative of the wider patient population.

“If you go down every branch of the policy tree, the numbers become very small,” Menon says. “It almost becomes like personalized medicine, because you can factor in age, race, gender—although gender didn’t matter in our study—PSA numbers, bilirubin numbers, etc.”

Menon has ongoing research projects aimed at improving health care through technology, at both the patient level (a la the prostate cancer study) and the ecosystem level (a la the CDSS study). One paper in progress focuses on Covid-19 and how the data-sets research scientists selected for their studies influenced their findings. Another looks at telemedicine’s effects on quality of care.

“My foray into health care began with my PhD dissertation, which was on IT in hospitals,” Menon says. “At that time, I was working primarily from a hospital administration point of view. As a business school researcher, it seemed logical to stay there. But as you come across more problems, and you read more, you realize that the patient is the center of everything, not the hospital.”

Topics

Costello Research Data Analytics

Costello Research ICT

Costello Research Health and Well-being at Work

Would you rather buy from a cuddly chatbot, or the “Lipstick King”?

Jennifer Anzaldi — Tue, 04 Mar 2025 18:03:48 +0000

Would you rather buy from a cuddly chatbot, or the “Lipstick King”? Jennifer Anzaldi Tue, 03/04/2025 - 13:03

In This Story

Si Xie

Historically, entertainment and advertising have worked as a tag team, taking turns soliciting attention from audiences. But our social-media age is blending the two into new, hybrid forms.

Witness livestream shopping, a seamless amalgam of e-commerce and entertainment. In place of one-way messaging delivered by polished pitchpeople, this model employs relatable influencers presenting products for online sale—and chatting with consumers—in real-time sessions that often last several hours.

Famously popular in China, livestream shopping is picking up steam in the United States. In June 2024, as an example, U.S. TikTok netted its first million-dollar livestream, courtesy of Texas-based brand Canvas Beauty. By 2026, live shopping may be responsible for as much as five percent of all e-commerce sales in the U.S., according to industry projections.

For global brands, this means a possible revenue explosion. But for information-systems scholars like Si Xie, assistant professor at Costello College of Business, the global rise of livestream shopping represents an unprecedented research opportunity.

Si Xie

“One of the most important elements of livestream shopping is the interaction,” Xie says. “Livestreams bring all potential buyers into the same virtual room, together with the influencer. People can see which products have been put in the online shopping cart, and which have been purchased.”

Her recently published paper in Production & Operations Management finds that the longer an individual product is showcased in a livestream featuring several different brands, the more revenue it will generate. Yet as product showcase duration goes up, overall revenue from the livestream goes down.

To reach their conclusions, the research team—including co-authors Siddhartha Sharma of Indiana University and Amit Mehra of University of Texas at Dallas—analyzed data from nearly 75,000 livestreams conducted in China during 2021.

For Xie, the findings point to a fundamental conflict between the incentives of livestreamers and the brands they promote. It is in the best interest of third-party influencers to move fairly rapidly between different types of products, but brands will want more airtime devoted to each one.

“People like variety,” Xie explains. “If I watch a livestream and all I see are shirts in different fabrics, I might feel there are not too many choices I can make. However, if you show me a shirt and then a pair of pants, I can make an outfit. There’s a higher probability of my making more purchases, and that’s in line with the third-party livestreamers’ incentives.”

One way to correct these misaligned incentives would be for brands to use the power of the purse to influence the influencers. In China, even the suggestion of such corrupting relationships has caused public scandal. In 2023, for example, top livestreamer Li Jiaqi (nicknamed “The Lipstick King” for his ability to sell beauty products) lost one million followers on social media after lashing out at an online commenter who complained about the high price of an eyebrow pencil made by Chinese cosmetics company Florasis. Li, Florasis’s most prominent brand ambassador, was excoriated for ostensibly putting his relationship with the brand above empathy for financially struggling consumers.

“People were saying, ‘you are trying to be defensive of the product because you get so much interest from selling that pencil’”, Xie says. “Therefore, Li’s credibility was really impaired.”

If Xie’s paper describes how human imperfections can jeopardize livestream shopping, could AI be the answer? Indeed, AI-powered animated chatbots — both paired with human influencers, and serving customers solo during off-peak sales hours — have become commonplace on China’s livestreams. For her PhD dissertation, Xie probed data from more than 70,000 livestreams in China and found that introducing an AI assistant boosted livestream sales by about 18%. But the effect steadily declined over time — and not because the novelty wore off. The rapidly improving algorithmic responses had the unintended consequence of shorter watch durations, which may have reduced impulse buying. Xie’s suggested remedy? “The owner of the gen-AI tools could modify the interaction between the virtual livestreamer and the audience to encourage more engagement, perhaps by adjusting the learning speed to ensure that the audience remains engaged for a longer period."

Xie also suggests that brands and channels replace humanoid avatars with cute, cuddly “mascots” that users just can’t bring themselves to click away from.

Xie says she’s working on future papers that tease insights out of livestream data. “One good thing about this new technology is that it promotes the user to buy using methods we can observe. Livestreamers sell general items like grocery items and clothing, as well as expensive stuff like cars and houses, and you can really see how people behave.”

Topics

Costello Research Innovation Strategy

Costello Research ICT

Costello Research Brand Management

George Mason professor furthers impact of telemedicine in Ukraine

Jennifer Anzaldi — Tue, 14 Jan 2025 22:39:16 +0000

George Mason professor furthers impact of telemedicine in Ukraine Jennifer Anzaldi Tue, 01/14/2025 - 17:39

Ukraine’s health care system has been hit hard amid the ongoing war. Power outages, staffing shortages, and the destruction of hospitals have added up to a drastic reduction in available care for the already-vulnerable population. In a desperate attempt to bridge the gap, Ukraine’s Ministry of Health opened the country to telehealth solutions from overseas. But will these prove to be a successful substitute for at least some necessary services, or turn out to be no better than a tech Band-Aid?

Answering that question is where Mariia Petryk, assistant professor of information systems and operations management at the Costello College of Business at 91��, comes in. In her spare time, she works as volunteer director of analytics for TeleHelp Ukraine (THU).

Mariia Petryk

Founded by a cross-disciplinary group of Stanford students shortly after the war’s inception, THU was designed to succeed where other telemedicine initiatives in crisis-affected areas have failed. The founders worked tirelessly to assemble an international volunteer network comprising medical professionals, translators, interpreters and administrative “health navigators.” Aware that medical consultations were only part of the patient journey, THU’s founders sought to address the entire continuum of care.

Petryk stresses that while the project originated at Stanford, the technical team included “people from Chicago, Boston, other California schools…some very active volunteers were in Australia, South Korea, Canada and other countries.”

Petryk, herself of Ukrainian descent, was honored to lend her data science expertise to this worthy project. As analytics director, she manages a dozen or so number-crunching volunteers who measured and documented THU’s impact upon Ukraine’s displaced population during the initiative’s first full year.

As Petryk explains, “The Russian invasion created a humanitarian crisis where a lot of people were internally displaced. And when people relocate to a new place, they don’t know where to go for health care. They also are at higher risk for many issues, including mental health problems. And they don’t know where to turn to treat chronic diseases they may have.”

THU’s primary focus during its first year was delivering much-needed services to this population of war-ravaged internal exiles.

Petryk’s analytical work gave rise to a recent case study of THU published in Journal of Global Health. The paper’s other lead author was Aditya Narayan, a Stanford medical student and THU’s director of implementation and evaluation.

Their findings describe some impressive early successes. THU facilitated more than 1,200 virtual patient appointments from May 2022 to May 2023 alone. Despite often-chaotic conditions, patient attendance rates were above 70 percent for nine of the 13 months studied. As the first year wore on, the THU team found ways to prevent no-shows—for example, employing the popular texting platform Viber to communicate with patients and assigning an individual health navigator to each patient.

Even more impressively, 96 percent of patients reported that their health complaints were at least partially resolved during their visit.

The paper argues that aspects of THU’s model could be adapted for use in other humanitarian contexts. In its initial growth phase, THU had access to advanced technological infrastructure and a wide network of medical providers, by dint of its academic origins. This implies that partnerships with academia could be critical to replicating THU’s success outside Ukraine.

Petryk remains proud of THU’s impact and her role in helping define it. “Based on actual appointments and how much that amount of care would cost at a hospital, THU delivered an estimated $1 million worth of services in its first 13 months,” she says.

Looking ahead to THU’s future, she says, “I can only wish to see this ‘start-up,’ as it were, go for the IPO.”

For more information and to explore volunteering opportunities, visit THU’s website.

In This Story

Mariia Petryk

Connect with the Costello College of Business

Topics

Costello Research Data Analytics

Costello Research ICT

Costello Research Health and Well-being at Work

Faculty and Staff News

Off the Clock

STEM outreach

Are U.S. ‘news deserts’ hothouses of corruption?

Jennifer Anzaldi — Tue, 19 Nov 2024 16:35:27 +0000

Are U.S. ‘news deserts’ hothouses of corruption? Jennifer Anzaldi Tue, 11/19/2024 - 11:35

The March 24, 2021 edition of neighborhood newspaper Northeast News, out of Kansas City, Missouri, contained a surprise for its 9,000 subscribers. Where the front-page news should have been, there was a big, blank white space. This was no printer’s error, but a last-ditch cry for help. After 89 years in operation, Northeast News had found itself on the brink of insolvency due to the loss of key advertisers amid the COVID pandemic. The empty front page was designed to remind the community of what it would lose if its only local paper went under.

The gambit went viral, prompting a flood of online donations that is keeping the paper afloat, for now. Ironically, Northeast News owes its existence to the very force that has fueled the more general decline of local journalism in America—the internet.

Brad Greenwood

As advertiser dollars migrated to Facebook and Google, the business model that supported local newspapers for generations came to the edge of collapse. Since 2004, more than 2,500 American newspapers have ceased publication—around one-quarter of the total. Overall newspaper circulation has declined by more than half since 1990.

To be sure, digital alternatives have rushed in to fill the gap, such as citizen-journalist websites, nonprofit news organs, partisan blogs, etc. So, the question represented by the blank front page of Northeast News resonates: What do communities lose when newspapers fold that online journalism startups haven’t (so far, at least) been able to replace?

In the past, industry observers and researchers have linked community newspaper closure to diminished civic trust and political participation, among other negative effects. New research from Brad Greenwood, the Maximus Corporate Partner Professor of Business at the Costello College of Business at 91��, builds on this discourse, finding evidence that when local papers topple, political corruption springs up in their wake.

Greenwood’s paper, coauthored by Ted Matherly of Tulane University, was published in MIS Quarterly.

The researchers focused on U.S. federal districts that lost a major daily newspaper during the years 1996 to 2019. They compared the number of corruption charges (bribery, embezzlement, fraud, etc.), defendants, and cases filed in district court before and after the newspaper closure. The results were striking: Overall, the disappearance of a newspaper delivered a 6.9% increase in charges, a 6.8% increase in the number of indicted defendants and a 7.4% increase in cases filed.

“We looked at federal charges for three reasons. First, the overwhelming amount of statutory enforcement occurs federally. Second, it gives us a uniform definition of what constitutes corruption across every domestic jurisdiction. Finally, and most importantly, federal conviction rates are over 90%,” Greenwood says. “They don’t charge people unless they have a good-faith belief they will prevail at trial.”

Moreover, post-newspaper corruption cases were more likely to go to trial as opposed to resolving in a plea deal, thus incurring greater public costs.

“In an age of misinformation, the solution is not rejecting the professional press, it is embracing it, and ensuring that well-trained and hard-working men and women have both the ability and venue to hold those in power to account."

Greenwood and Matherly also examined whether digital-era upstarts were adequate substitutes for newspapers, in terms of curtailing corruption. They tracked 352 such websites, and found they had no impact on the number of charges, defendants or cases in the districts concerned.

“While it’s hard to say precisely why we don’t see an effect from online news, there are several candidate explanations. Not only do citizen journalists lack the standing and training to tackle questions of public corruption and elevate discourse in the public square, but many of these sites aren’t even legitimate news vendors,” says Greenwood, referencing what are commonly referred to as “pink slime websites.”

Greenwood goes on to suggest that the corruption-preventing power of the defunct papers came not necessarily from journalistic acumen, but rather from the ability to elevate the actions bad actors had taken in public discourse, a process journalism researchers refer to as agenda setting.

Whatever the cause, the ramifications for society are very real. In the Northern District of Illinois alone, corruption-related cases involving more than 1,700 officials cost taxpayers a staggering $550 million per year from 1976 to 2012. The coffers of communities that lose newspapers may suffer more than most, since these cases tend to end up in expensive courtroom proceedings rather than plea deals.

Further, the study only looks at corrupt officials who got caught. Presumably, there are many more whose corruption went unpunished.

All told, these findings suggest that community newspapers should not be regarded as just another business model ill-adapted to digital disruption that should be allowed to fail. Their demise comes at significant public cost, financial and otherwise. “In an age of misinformation, the solution is not rejecting the professional press, it is embracing it, and ensuring that well-trained and hard-working men and women have both the ability and venue to hold those in power to account,” Greenwood says.

In This Story

Brad Greenwood

Topics

Costello Research Business Model Innovation