A market-research vendor that sold fabricated survey data to academic researchers for roughly a decade has produced its first academic retraction, and prosecutors’ case against the people who ran it is still working its way through federal court. Retraction Watch reported on August 6, 2026 that studies drawing on panel data sold by Op4G (“Opinions for Good”) and its successor firm Slice MR reached research teams at Northwestern University, the University of Maryland, Wake Forest University, the University of North Carolina at Chapel Hill, the University of Chicago, and Northwell Health. One paper has now been retracted, two studies are under review for correction, and at least one NIH-funded project had to re-collect its data from scratch before publication.
What Op4G and Slice MR did
According to the indictment unsealed in April 2025 in the U.S. District Court for the District of New Hampshire, Op4G — founded in 2010 and based in New Hampshire — sold survey-panel data to market-research and academic clients. Starting in 2014, according to prosecutors, staff including chief operating officer Frank Hayden and Daniel Harriman began recruiting paid fake respondents, internally called “ants,” based in the United States, Serbia, India, and Brazil, to complete surveys for a nominal fee. Legitimate responses were then mixed with the fabricated ones in a process the defendants themselves reportedly called “scrambling the eggs.” In 2018, part of the operation continued under a second company, Slice MR, based in Illinois.
Prosecutors allege the two firms billed clients more than $10 million for data tainted by fabrication between 2014 and early 2025, and that by the later years of the scheme, fabricated responses accounted for roughly 90% of the two companies’ revenue. The fraud came to light after Kristen Tripphahn, who joined Op4G as chief operating officer in February 2022, discovered the practice internally and became the case’s whistleblower. “I discovered 6 months in that nothing could have been further from the truth,” she has said of the company’s actual data-collection practices.
The criminal case so far
Eight people were indicted on one count each of conspiracy to commit wire fraud: Frank Hayden, Daniel Harriman, Frank Nappo, Ryan Stoudt, Katarina Grubljesic, Strahinja Grubljesic, Archie Ignacio, and Arvind Iyer (also known as S. Aravindan). Three have since pleaded guilty — Hayden, Harriman, and Stoudt — with sentencing proceeding before Judge Landya B. McCafferty; the remaining five defendants’ cases had not been resolved as of Retraction Watch’s August 6 report. Readers following the case should treat anything not yet reflected in a plea, verdict, or sentencing order as an allegation, not a finding, against the defendants who have not entered a plea.
Which studies have been affected
No institution or publisher has released a comprehensive count of how many papers used tainted Op4G or Slice MR data — say that plainly rather than estimate it. Three concrete cases have surfaced publicly so far:
- Retracted. The Journal of Adolescent and Young Adult Oncology retracted a 2023 study led by Allison Lazard of UNC Chapel Hill on social media’s role in supporting young adults with cancer. The retraction notice states the authors “conclusively infer that fraudulent data was provided to them” and that they themselves requested the retraction in June 2026.
- Under review. Stacey Tessler Lindau at the University of Chicago used Op4G panel data, drawn from roughly 3,200 respondents, for two COVID-19 pandemic studies. She has said that “if any retractions or corrections associated with the use of that data are appropriate, those steps will be taken,” without yet announcing a specific outcome.
- Re-collected before publication. David Victorson at Northwestern used Op4G data, also drawn from about 3,200 respondents, for an NIH-funded study developing a mindfulness metric. Victorson learned of the fraud from the April 2025 DOJ press release, at which point his team had a paper close to submission. “We were just about to publish our primary paper based on those data when we found out about this, and we stopped everything,” he has said. The team re-collected and re-analyzed the data before publishing; the study’s primary findings held.
The named researchers above are described in every source reviewed for this article as clients defrauded by their data vendor, not as subjects of a misconduct finding — that distinction matters and this article preserves it throughout.
Why this is a research-administration problem, not just a publishing one
Purchased survey panels sit in a governance gap. Institutional review board review and data provenance documentation are generally built around what a research team does with data after it arrives, not around verifying that a third-party vendor’s respondents were real people who actually answered the questions asked. A data management plan can describe storage, access, and retention for panel data in detail without ever addressing how the data was originally collected or how its authenticity was checked upstream, because DMP templates were not written with vendor fraud as a threat model. That is exactly the gap the Op4G case exposes: the fabrication happened upstream of every control a typical human-subjects protocol or DMP actually reviews.
Institutions procuring panel or survey data from a commercial vendor should treat data-provenance terms as a contracting issue, not an afterthought bundled into a standard purchase order. At minimum, a provenance clause for purchased panel data should require the vendor to disclose its respondent-recruitment and quality-control methodology in writing, name the specific fraud-detection checks it runs (attention checks, duplicate-device or duplicate-IP detection, response-time screening, geolocation verification), retain and make available on request the raw metadata needed to audit a sample of responses, and provide contractual notice if the vendor becomes aware of, or is charged with, data integrity problems affecting delivered data.
For a principal investigator who learns mid-study that a vendor has been indicted or is under investigation, David Victorson’s response is a workable template: stop before publication, re-collect the affected data through a verified channel, re-run the analysis, and only then decide whether to proceed, correct, or retract based on whether the original conclusions hold. Institutions should also flag any already-published work that relied on the same vendor and route it to the office responsible for retraction and correction decisions, since — as this case shows — the review can surface work published years before a vendor’s fraud becomes public.
What to watch next
Two things keep this story open rather than settled. Five of the eight indicted defendants have not yet entered a plea, so further guilty pleas, a trial, or additional charges remain possible. And the retraction and correction process for research built on Op4G and Slice MR data has only produced one retraction so far, against an unknown total number of affected publications — meaning further retraction or correction notices citing this case are likely as more institutions complete their own reviews.
