Freakonomics Radio

572. Why Is There So Much Fraud in Academia?

Thu Jan 11 2024

research misconductacademic fraudpublic healthtransparencyreproducibilityquestionable practicesfraudulent datareplication failurelegal battlescientific inquirysocial science research

Description

The episode discusses research misconduct, academic fraud, and the impact on public health and academic standards. It explores the manipulation of data for desired findings, replication failure, and the legal consequences of academic fraud. The podcast raises concerns about the chilling effect on scientific inquiry and highlights the positive influences of social science research. The conversation will continue in the next episode, covering the financial aspect of research paper mills.

Insights

Research Misconduct and Fraud

Academic fraud is a widespread issue in academia, with high stakes due to potential harmful consequences in areas such as medicine and public policy.

Impact on Public Health

False ideas in academic research, such as the Wakefield scandal linking vaccines to autism, can have damaging effects on public health.

Transparency and Reproducibility

Transparency and reproducibility are crucial in scientific research to ensure accuracy and prevent fraud.

Questionable Practices in Research

Competition for attention and advancement in academia can lead to dishonest behavior and questionable research practices.

Fraudulent Data and Replication Failure

Fraudulent data and replication failure undermine trust in scientific findings and raise questions about scientific rigor.

Legal Battle and Scientific Inquiry

The legal battle surrounding academic fraud raises concerns about the chilling effect on scientific inquiry and the need for accountability.

Positive Influences of Social Science Research

Social science research has had positive impacts on various areas, including health, exercise, and financial planning.

Chapters

Research Misconduct and Academic Fraud
Impact on Public Health and Academic Standards
Transparency, Fraud, and Academic Fields
Manipulating Data for Desired Findings
Fraudulent Data and Replication Failure
Questionable Practices and Data Fabrication
Unreliable Findings and Fraudulent Data
Replication Failure and Fraudulent Data
Academic Fraud and Legal Consequences
Legal Battle and Implications for Scientific Inquiry
Positive Influences of Social Science Research

Summary

Transcript

Research Misconduct and Academic Fraud

00:06 - 08:12

Francesca Gino, a renowned academic in the field of behavioral science, was suspended without pay by Harvard Business School after being found guilty of research misconduct. Gino's reputation as an expert in organizational behavior and decision science was tarnished by the investigation.
The whistleblowers who exposed Gino also provided evidence of data fraud by another prominent behavioral scientist, Dan Ariely of Duke University.
Universities tend to downplay charges against their superstar professors to protect their own reputation.
Academic fraud is not limited to these two cases and is a widespread issue in academia.
Cheating and fraud have always existed in scientific research due to human motivations and biases.
The Center for Open Science aims to improve the integrity of scientific research through funding from various sources.
The stakes are high as compromised research can lead to harmful consequences in areas such as medicine and public policy.

Impact on Public Health and Academic Standards

07:50 - 15:22

The Wakefield scandal, which linked vaccines to autism, has had a damaging impact on public health and people's beliefs about autism and vaccines.
False ideas in academic research can lead to waste and hinder progress.
Universities should set an example for their students by upholding the standards of academic research.
Academic research should be accurate and free from personal or financial interests.
The Center for Open Science aims to promote transparency, rigor, and reproducibility in scientific research.
The academic reward system discourages researchers from sharing their data due to concerns about career advancement.
The Reproducibility Project found that less than half of the findings in psychology and cancer biology successfully replicated.
Many studies conducted at elite universities may not be producing science of lasting value.
Fraud is a corrosive element in the scientific system that undermines trust and transparency.

Transparency, Fraud, and Academic Fields

15:02 - 22:41

Transparency is important in scholarship, allowing others to examine evidence and interpretations.
The case of Joachim Bolt, a German anesthesiologist with almost 200 retracted papers, highlights the impact of faulty research on people's lives.
Fraud or sloppiness can occur in any academic field, but some gain more attention due to their interesting findings.
Social psychology receives a lot of attention because it addresses social problems within the scientific system.
The competition for attention and advancement in academia has increased, potentially leading to dishonest behavior.
Economics research tends to have fewer instances of shady research due to robustness checks, aggressive debate, and working with publicly available big data sets.
Cheating in research undermines the efforts of honest researchers and may require more invasive measures to address, such as interrogating suspicious papers and making public accusations of fraud.
Whistleblowers at the University of Pennsylvania run a blog called Data Calada and focus on examining research methodology.
Intuition sometimes overrides findings in academic research, leading to skepticism when results don't align with preconceived beliefs.

Manipulating Data for Desired Findings

22:15 - 29:12

The podcast discusses a study conducted by the Dada Colada team to demonstrate how easy it is to produce false evidence.
They chose to show that listening to a song by the Beatles, "When I'm 64," can make people younger.
The team manipulated and cherry-picked data to support their desired finding that listening to the song lowers your age by a year and a half.
The study was published in Psychological Science, but received mixed reactions from the field.
The Dada Colada team started their blog as a way to share shorter articles without waiting for the lengthy review process of journals.
"P" hacking involves selectively reporting certain measurements or dropping conditions from studies for self-serving purposes.
Convenient errors, like typos that favor expected results, were also highlighted as potential issues before actual fraud occurs.

Fraudulent Data and Replication Failure

28:56 - 36:10

The likelihood of fraud increases when certain conditions are present, such as dropping measures or changing participant answers.
Anomalies in data sets can indicate fraudulent activity.
Approximately 5% of academic articles may contain fraudulent data.
Fraud is less likely to be prominent in high-profile academic journals compared to lower-tier ones.
Data collata investigators have identified evidence of fraud in influential research papers.
A paper published in the Proceedings of the National Academy of Sciences claimed that signing at the beginning of a form promotes honesty.
The central finding of this paper gained popularity and was implemented by various institutions.
The involvement of renowned psychologists and evidence of non-replication raised suspicions about the original paper's validity.
Max Bayserman, a professor at Harvard Business School, became a co-author on the problematic research paper due to mentorship reasons.
The "Signing at the Top" paper originated from two separate research projects.

Questionable Practices and Data Fabrication

35:47 - 43:17

The paper started as two separate research projects, one about the effect of signing a document before filling it out on truth-telling and another about honesty pledges.
The studies were conducted in a laboratory setting at the University of North Carolina at Chapel Hill.
Research subjects in the experiments solved puzzles for financial rewards and then filled out a form to report their earnings. Some subjects were asked to sign an honesty pledge at the top of the form, while others signed at the bottom.
The data from these experiments showed that participants who signed at the top were more truthful about their earnings.
Francesca Gino played a significant role in collecting the data, but Lisa Shoe and Bayserman had little involvement in data collection.
Bayserman discusses how senior colleagues like him often do not interact directly with the original data but rely on junior colleagues for analysis.
The original paper was rejected by multiple journals, prompting researchers to seek additional real-world results to support their argument.
They collaborated with Dan Ariali, who had field experiment results from an insurance company showing that customers who signed mileage statements at the top reported driving more miles than those who signed at the bottom.
Bayserman noticed discrepancies in Ariali's data regarding average mileage driven per year and questioned its accuracy.
Despite seeking clarification, satisfactory answers were not provided.

Unreliable Findings and Fraudulent Data

42:53 - 50:24

People have driven many miles for the study, but it is unclear why.
The presenter of the study was concerned about the lack of clarity on the mileage issue.
The presenter meets Nina, who is supposedly involved in the insurance study.
Nina explains that there may have been a longer period between time one and time two for assessing the number of miles driven, making the study noisier.
The presentation in the paper is corrected and published.
The effect of signing tax forms at the top instead of the bottom leads to millions or billions of extra dollars flowing to tax agencies.
Stuart Baserman emails the presenter about getting people to tell the truth online and they develop a relationship. They are also fifth cousins.
The presenter becomes a consultant for Slice Insurance based on their findings from the 2012 paper.
In 2016, Max Bayserman tests whether signing at the top would be effective in an online setting but finds no effect.
There were concerns about a placebo effect influencing results, but it was not widely known at that time.

Replication Failure and Fraudulent Data

50:07 - 57:43

The podcast discusses a series of failed replications of an academic effect related to honesty in online communication.
The original paper from 2012 had significant effects across three different studies, but subsequent replications failed to find any effect.
A large-scale replication was conducted with more research subjects, but still no effect was found.
One reason for the replication failure could be the use of small sample sizes in original studies, which are more likely to produce skewed results.
The researchers felt a moral obligation to correct the record and published a follow-up paper stating that signing at the beginning or end does not decrease dishonesty.
The original authors did not retract the original paper, leading to questions about scientific rigor and fraud.
Max Bayserman, a senior researcher, received evidence from the data collada team showing fraud in one of his co-authored papers on insurance.
Suspicious data included a histogram showing an unusual distribution of miles driven by study participants.
New information from the insurance company confirmed that the published data differed significantly from what they had provided.
Dan Arielli, one of the original authors, acknowledged that the underlying data for their study was dishonest but couldn't determine exactly what went wrong.

Academic Fraud and Legal Consequences

57:15 - 1:05:00

Arielli, one of the researchers involved in a study about dishonesty, admitted that the data underlying the study was dishonest.
The co-authors of the study participated in review sessions and felt satisfied with the answers they received regarding the data.
The PNAS journal retracted the original paper.
Data Colada found evidence of data fabrication in one of Francesca Gino's lab studies and serious problems with her data in general.
Max Baserman, who had a close relationship with Francesca Gino, was shown evidence of data fabrication and declined to take it to Harvard.
Data Colada received a tip from a graduate student and another anonymous researcher suggesting that many more Gino-authored papers contain fake data.
Harvard University placed Francesca Gino on administrative leave and recommended retracting four papers or amending previously retracted ones.
The accuracy of Data Colada's analysis is highly confident according to Yuri Simonsson.
Francesca Gino sued Harvard and Data Colada for defamation, claiming she did not commit academic fraud and alleging discrimination based on sex.
Some questioned why Gino's punishment was severe while others faced less immediate discipline for plagiarism, including Harvard President Claudine Gay.

Legal Battle and Implications for Scientific Inquiry

1:04:40 - 1:11:54

Three individuals are being sued for $25 million, and they have set up a GoFundMe campaign to cover their legal expenses.
The lawsuit has raised concerns about the chilling effect on scientific inquiry and criticism.
Incentives within academia can lead researchers to exaggerate their work and oversell their findings.
The replication crisis in psychology has led to intense self-examination within the field.
There is a need for soul searching and accountability among researchers to prevent similar issues in the future.
Despite the challenges, social science research has had positive impacts on areas such as health, exercise, and financial planning.

Positive Influences of Social Science Research

1:11:24 - 1:14:02

Social science has many positive influences, such as promoting healthier eating habits, increased exercise, and better retirement savings.
Social scientists have been successful in helping people move in the right direction through credible research.
The credibility of social science research could be undermined if fraud becomes the dominant narrative.
In the next episode, the conversation will continue from different angles, including the financial aspect of research paper mills.
Research paper mills charge hundreds to thousands of dollars per paper and publish tens of thousands or more papers per year.
Freakonomics Radio is produced by Stitcher and Renbud Radio and has an archive dating back to 2010. Transcripts and show notes are available on Freakonomics.com.