In relation to “Why don’t the raw data match what was reported in a chemistry paper?”, published in Retraction Watch, relating to academic fraud, Neil Remington Abramson commented as follows:
There is an avalanche of papers being retracted from scientific journals – even top journals – because the reported findings are inconsistent with the data collected. Authors may be falsifying their evidence to produce the kinds of excellent and significant results needed to get top journal hits. Without these hits, a scholar who has spent 5-7 years getting his/Her PhD and 7-8 years on probation as an untenured faculty member, will not obtain tenure. They will lose their professorship and possibly the value of all those years.
A 2015 study of 100 research articles published in top psychology journals found that the majority’s findings could not be replicated when similar methods were employed. Then the study criticising the studies had its own problem with replication. The initial study conclusions: Either results reported as significant were no longer significant, or not nearly as significant as originally reported. However, maybe the initial study’s findings aren’t that significant.
“Significant” means that a result is so extreme that it could not have happened by chance. P >.001 means it could only happen by chance less than 1 in a thousand times. P > .01 or .05 means less than 1 in a hundred, or 5 in a hundred chances.
Less than a thousand is far more likely to interest a top journal than less than a hundred. The difference is only one “0” and could make the difference between a long successful professorial career, or 12-15 wasted years pursuing a career that was not to be.
You can understand why someone would be tempted do it. You shouldn’t do it however, and we profs are trained not to. It’s unethical and it perverts and undermines science. Anyone caught doing it should be terminated, if it’s proven.
Though proven on the balance or probabilities or beyond reasonable doubt? That’s significant, as well.