Even the best poker players have "behavioral language" that may reveal their thoughts when playing cards when they bluff with one hand. The same is true for scientists who fake
. When they are forging data, they will also try to cover up the falsehood in their works through some expression.
A team of researchers at Stanford University have invented a method to identify these written clues.
, a research published in the Journal of Language and Social Psychology, can help us identify fake studies in a unique way.
In order to understand the way liars lie, the academic community has already had a considerable number of research results. These studies suggest that liars tend to express more negative emotions and use fewer first-person pronouns.
(I would like to remind you that the editor recently opened a video account to talk about the postgraduate entrance examination. I plan to appear in the future. Please pay attention and like it.)
In the field of financial reporting, compared with accurate reports, those false reports usually have a higher degree of confusing language, which are wordings designed to disperse or hide false data.
To find out whether there is a similar pattern in the scientific community, Stanford Communications professor Jeff Hancock and graduate student David Markowitz searched the archives of PubMed, a life science journal database, from 1973 to 2013, and found the papers that were withdrawn for research.
They identified 253 papers mainly from biomedical journals that were withdrawn because of the discovery of fraud.
Then, they compared these papers with unretracted papers covering the same topic in the same journal and the same year of publication.
Then, they rate the degree of fraud in each paper with a customized "obfuscation index" that divides the degree to which the authors try to mask false conclusions.
This degree of division comprehensively considers the causal words, abstract language, "jnaparalleries", positive emotional terms and legibility in the article.
The paper's lead author Markowitz said: "The underlying idea behind confusion is to get confused. Scientists who falsify data know that they are doing wrongdoing and do not want to be caught. Therefore, a strategy to avoid this may be to blur some of the contents of the article. We believe that language can be one of the many variables that distinguish pseudoscience from real science." The
study shows that papers that are withdrawn due to falsification have a significantly higher "confusion index" than papers that are withdrawn due to other situations. In addition, in fake papers, the rate of certain "javascript" appears 1.5% higher than that of unretracted papers.
Markowitz said: "Compared with papers that have not been withdrawn, there are about 60 additional words similar to 'jnapark' per article. This is a big number."
researchers say that scientists may falsify data for various reasons. Previous research pointed out that a pressure to “publish or perish” may inspire researchers to modify their research results or simply fake them.
In this study, the researchers found the direct correlation between the authors’ writing style and their attempts to cover up the purpose of lies by manipulating language.
For example, a fake author may use less positive sentiment terms to curb praise for the resulting data because they are afraid to cause investigations.
Researchers said that based on the magazine's restrictions on fuzzy language, a computer system can be developed in the future to identify newly submitted papers so that editors can review the article more strictly before it is published.
But the author warns that this approach may not be feasible at present given the false positive rate.
Hancock said: "Scientific fraud is getting more and more attention in the academic community, and tools that automatically identify fraud may be useful. But before considering this approach, more research is needed."
"Obviously, the problem that needs to be solved is that there is a very high error rate in this.However, science needs to be based on trust, and the introduction of 'false detection' tools during publication may undermine this trust. "He said.
shares the latest scientific and technological information and publish cutting-edge academic updates! Everything is in Mumuxili~ Serving scientific research and helping technology! Follow the WeChat public account: Mumuxilinj, more exciting content, news information, and dry resources are waiting for you!
Special statement: This article is published only for the need to disseminate information and does not represent the views of this public account; if other media, websites or individuals reprint from this public account, please apply to the original author and bear legal responsibilities such as copyright.
Click the end of the article Learn more , and receive a million gift!
