A portion of this topic was covered in the April 2018 blog, but enough additional material has surfaced that this information needs to be covered again. The reproducibility issue covered in April is one thing. The ability to slant results through the use of statistics [Ref. #1] has been known for a long time. The omission of data points that don’t support the conclusions is another method. It is always possible to come up with an argument that supports changing of results, but is that proper?
The National Institutes of Health Clinical Center Department of Bioethics has a brochure [Ref.2], which has 7 key steps for ethical development of research. Two of these are Scientific Validity and Independent Review. The impact on society is a key consideration for the overall research. Medical research is directed at saving lives and improving the quality of life for impacted people. Consider the following situations:
Professor Brian Wansink of Cornel University was considered a leading researcher in eating behavior. He resigned earlier this year due to findings that he misreported research data, employed problematic statistical techniques, did not properly document and retain research results. The main contention of his statistical work was that he employed one technique (p-hacking) that involves running statistical analyses until statistically significant results are obtained and the other (HARKing) is hypothesizing after the results are known. [Ref. 3]
The head line reads: “Harvard Call for Retraction of Dozens of Studies by Noted Cardiac Researcher.” [Ref.. 4] Dr. Piero Anversa published results suggesting that damaged heart muscle could be regenerated with stem cells. Although his work could not be replicated by independent researchers. There were numerous awards for clinical trials. The questioning of his work resulted in more than 30 published papers that were in question. The result was that an entire field of study developed by Dr. Anversa is called into question. His institution, Brigham and Women’s Hospital – A Harvard medical School paid #10 million for research fraud. [Ref. 5]
Duke University had a researcher in the lab of a prominent pulmonary scientist arrested on charges of embezzlement. [Ref. 6] The investigation turned up some unusual things. The end result was the 15 or the scientist’s papers were redacted. It was claimed that the research in these papers had enabled Duke University to obtain over $200 million in grants.
Unfortunately, these are not isolated cases. The onlineuniversities web site provides more details of the “10 Greatest Case of Fraud in university Research.” [Ref. 7] It is worth a quick skim to see the areas of research and the impact on people. Remember these sentences in the second paragraph? The impact on society is a key consideration for the overall research. Medical research is directed at saving lives and improving the quality of life for impacted people.
Other areas of scientific research are not immune from “interesting” shenanigans. A recent article in the Washington Post [Ref. 8] contains a claim by a scientist that the oceanographic study recently released contains errors that increased the possibility of the project results. The response was that the scientists were work quickly to create the report and may have included inadvertent mistakes. Understandable. However, the National Oceanic and Atmospheric Administration (NOAA) has refused to provide the research data, notes, etc., to Congress. [Ref. 9] As a federal agency, NOAA receives its budget from Congress and Congress has oversight responsibility.
The last reference has a number of interesting observations. [Ref, 10] A key point is that developing crises results in the need to investigate and understand the cause and impact of the developing crises. There is a reference from Al Gore’s book quoting Upton Sinclair. It is difficult to get a man to understand something when his salary depends upon his not understanding it.”
If we can’t believe that scientific research is driven by facts, hypothesis development and testing, and then valid conclusion based on reproducible experiments, how can we trust the actions presented as needed by the results?
- How to Lie with Statistics, Darrell Huff, ISBN-13: 978-0393310726, ISBN-10: 0393310728
- Scientists acknowledge errors in Study of Oceans, Austin American Statesman, Thursday, November 15, 2018, page A8
- Onward, climate soldiers, https://www.washingtontimes.com/news/2018/nov/13/science-loses-when-a-system-of-penalties-and-rewar/