6.0 KiB
| title | chunk | source | category | tags | date_saved | instance |
|---|---|---|---|---|---|---|
| Invalid science | 1/3 | https://en.wikipedia.org/wiki/Invalid_science | reference | science, encyclopedia | 2026-05-05T04:28:14.614270+00:00 | kb-cron |
Invalid science consists of scientific claims based on experiments that cannot be reproduced or that are contradicted by experiments that can be reproduced. Recent analyses indicate that the proportion of retracted claims in the scientific literature is steadily increasing. The number of retractions has grown tenfold over the past decade, but they still make up approximately 0.2% of the 1.4 million papers published annually in scholarly journals. The U.S. Office of Research Integrity (ORI) investigates scientific misconduct.
== Incidence == Science magazine ranked first for the number of articles retracted at 70, just edging out PNAS, which retracted 69. 32 of Science's retractions were due to fraud or suspected fraud, and 37 to error. A subsequent "retraction index" indicated that journals with relatively high impact factors, such as Science, Nature and Cell, had a higher rate of retractions. Under 0.1% of papers in PubMed had were retracted of more than 25 million papers going back to the 1940s. The fraction of retracted papers due to scientific misconduct was estimated at two-thirds, according to studies of 2,047 papers published since 1977. Misconduct included fraud and plagiarism. Another one-fifth were retracted because of mistakes, and the rest were pulled for unknown or other reasons. A separate study analyzed 432 claims of genetic links for various health risks that vary between men and women. Only one of these claims proved to be consistently reproducible. Another meta review, found that of the 49 most-cited clinical research studies published between 1990 and 2003, more than 40% of them were later shown to be either totally wrong or significantly incorrect.
=== Biological sciences === In 2012, biotech firm Amgen was able to reproduce just six of 53 important studies in cancer research. Earlier, a group at Bayer, a drug company, successfully repeated only one fourth of 67 important papers. From 2000 to 2010, roughly 80,000 patients took part in clinical trials based on research that was later retracted because of mistakes or improprieties.
=== Paleontology === Nathan Myhrvold failed repeatedly to replicate the findings of several papers on dinosaur growth. Dinosaurs added a layer to their bones each year. Tyrannosaurus rex was thought to have increased in size by more than 700 kg a year, until Mhyrvold showed that this was a factor of 2 too large. In 4 of 12 papers he examined, the original data had been lost. In three, the statistics were correct, while three had serious errors that invalidated their conclusions. Two papers mistakenly relied on data from these three. He discovered that some of the paper's graphs did not reflect the data. In one case, he found that only four of nine points on the graph came from data cited in the paper.
=== Major retractions === Torcetrapib was originally hyped as a drug that could block a protein that converts HDL cholesterol into LDL with the potential to "redefine cardiovascular treatment". One clinical trial showed that the drug could increase HDL and decrease LDL. Two days after Pfizer announced its plans for the drug, it ended the Phase III clinical trial due to higher rates of chest pain and heart failure and a 60% increase in overall mortality. Pfizer had invested more than $1 billion in developing the drug. An in-depth review of the most highly cited biomarkers (whose presence are used to infer illness and measure treatment effects) claimed that 83% of supposed correlations became significantly weaker in subsequent studies. Homocysteine is an amino acid whose levels correlated with heart disease. However, a 2010 study showed that lowering homocysteine by nearly 30% had no effect on heart attack or stroke.
=== Priming === Priming studies claim that decisions can be influenced by apparently irrelevant events that a subject witnesses just before making a choice. Nobel Prize-winner Daniel Kahneman alleges that much of it is poorly founded. Researchers have been unable to replicate some of the more widely cited examples. A paper in PLoS ONE reported that nine separate experiments could not reproduce a study purporting to show that thinking about a professor before taking an intelligence test leads to a higher score than imagining a football hooligan. A further systematic replication involving 40 different labs around the world did not replicate the main finding. However, this latter systematic replication showed that participants who did not think there was a relation between thinking about a hooligan or a professor were significantly more susceptible to the priming manipulation.
== Potential causes ==
=== Competition === In the 1950s, when academic research accelerated during the Cold War, the total number of scientists was a few hundred thousand. As of 2023, there are now a total estimated 8 million researchers worldwide. The number of research jobs has not matched this increase. Every year six new PhDs compete for every academic post. Replicating other researcher's results is not perceived to be valuable. The struggle to compete encourages exaggeration of findings and biased data selection. A recent survey found that one in three researchers knows of a colleague who has at least somewhat distorted their results.
=== Publication bias === Major journals reject in excess of 90% of submitted manuscripts and tend to favor the most dramatic claims. The statistical measures that researchers use to test their claims allow a fraction of false claims to appear valid. Invalid claims are more likely to be dramatic (because they are false.) Without replication, such errors are less likely to be caught. Conversely, failures to prove a hypothesis are rarely even offered for publication. "Negative results" now account for only 14% of published papers, down from 30% in 1990. Knowledge of what is not true is as important as of what is true.