What Is It?
The more colloquial way to describe fabrication of data is lying. It is one instance of scientific misconduct that is also, without question, a violation of the moral ethics of decent people. In this ethical violation, researchers quite literally make up data, claiming experiments were carried out when they were not or significantly altering the results they do obtain so that they fit the pre-experimental hypothesis or previous studies. This ethical violation is among the most difficult to catch. This is because data fabrication can only be caught if another scientist attempts to repeat or otherwise use the research that was fabricated. This ethical violation is also perhaps the most damaging of them allânot just because it is also, undeniably, morally wrong, but because it also has the unfortunate consequence of leading other researchers down an incorrect and potentially impossible path trying to repeat and/or use the fabricated results. This then has many negative effects on other researchersâ careers resulting in wasted time and research funds. In these days of intense pressure to publish and obtain grants in times of tight financial stress, losing time and/or money could result in grants being terminated or denied funding and/or in the destruction of a new faculty memberâs chances of being awarded tenure or promotion. Consequently, fabrication of data violations is often met with severe repercussions such as termination from a position or bans from applying for federal grants. Even with such serious threats hanging overhead, many people still succumb to the temptation for reasons that will be touched upon below.
Consider This Hypothetical Scenario
FALSIFYING DATA
A researcher is searching for new planets using an infrared telescope, measuring how the heat signature of the stars decreases as the planets pass between our relative position and the starsâ. It is a total failure, but they create a table of results that describes the effect being observed and write it up describing the identification of planets previously discovered, claiming it is proof the method works.
FABRICATING DATA
A researcher believes that planets can be discovered using an infrared telescope. The researcher claims that as the planets pass between Earth and the star, they absorb enough heat to be measured. The researcher calculates how much heat would be absorbed with Mercury and Venus as they pass between Earth and the Sun and then claim that using the infrared telescope, and is able to detect when the planets traverse the Sun between 25% and 75% with the maximum calculated absorption at 50%; meanwhile, no experiments were carried out.
Also falling into this category is alteration of results (especially spectroscopic results such as NMR) to make the data appear more like that which was expected or desired, or to make the product of a chemical reaction appear purer. Although some may argue that âItâs no big deal, I only photo-shopped out the solvent,â it is still scientific misconduct. It is most unfortunate that this is virtually impossible to catch. Many would argue that this is not as bad as quoting a crude yield of 96% while neglecting to report that the pure yield is 34%, and perhaps thatâs true though I cannot agree. These are, without question, bona fide fabrications of data. With the advent of more and more advanced computer programs, this type of data fabrication is becoming easier with each passing year. In fact, somewhat recently, the Journal of Biological Chemistry announced its adoption of the Journal of Cell Biologyâs policy because this has become a more prevalent issue that reviewers and readers must be more cognizant of than ever before.1 The policy reads: âNo specific feature within an image may be enhanced, obscured, moved, removed, or introduced. The grouping of images from different parts of the same gel, or from different gels, fields, or exposures must be made explicit by the arrangement of the figure (e.g. using dividing lines) and in the text of the figure legend. Adjustments of brightness, contrast, or color balance are acceptable if they are applied to the whole image and as long as they do not obscure or eliminate any information present in the original, non-linear adjustments (e.g. changes to gamma settings) must be disclosed in the figure legend.â The Journal goes on to list the procedure that they feel will ensure the prevention/detection of such misconduct. Their comments conclude with the following: âAfter due process involving the JBC editors, editorial staff and ASBMB Publications Committee, papers found to contain inappropriately manipulated images will be rejected or withdrawn and the matter referred to institutional officers.â
Fabrication of data is significantly different from the data that is accused of being erroneous but not fabricated. I would argue that the latter of these cases are not an example of bad science or bad ethics but an example of scientific progress. Take as a brief example now the ancient belief that Earth was at the center of the universe. Based upon the evidence that was collected, this was indeed a logical conclusion for ancient humans to make. Modern evidence refutes this however, and we now know this to be incorrect. This, for certain, doesnât make ancient manâs approach toward this conclusion unethical. It is also not bad science. They took the data they had and made what they believed to be the most logical conclusion. As science has progressed, weâve become capable of not only proving that Earth is not at the center of the universe but also not even at the center of the galaxy or even our stellar neighborhood, the solar system. This is clearly a case of scientific progress and our ability to both collect and interpret data. Examples like this abound especially in medicine and are neither bad science nor scientific misconduct.
Why Does It Happen?
Why do people fabricate their data? Well, this one should be easy, even for someone who has nothing to do with any scientific field to answer. Many reasons, in fact, should come to mind instantaneously. For one, it is exceedingly difficult to catch, making it (potentially) very easy to get away with. Second, youâll very rarely, if ever, get a grant or publish a manuscript based on poor results, and without these (grants and publications), youâre not going to keep your job very long and you will certainly have a difficult time earning tenure, a promotion, or a raise. Excuses for a pharmaceutical company to fabricate data are even clearerâmany millions or even billions of dollars hang in the balance. The incentive to display good results is clear; your career and livelihood depend on it. The greed to be (or appear to be, more accurately) the best is a very tempting thing to some people.
How Is It Caught?
This is one of the worst forms of ethical misconduct and one of the hardest to catch. It is simply impossible to catch this via the peer-review process (which will be discussed in detail later), as that would require a reviewer to check every claim.2 With the enormous volume of work and increasing specialization of research, such a widespread effort is nothing short of unreasonable. Wholesale fabrication of data is impossible to catch before publication. Usually, instead, this violation is learned the hard way by an innocent researcher trying to use or further develop the fabricated results.
One way to potentially catch the digital alteration of results may be to take advantage of the improving computer technology that the perpetrators exploit and have raw data files (that have the appropriate time and date stamps) sent to reviewers. The reviewer could then use the appropriate software to recreate the figures and perform a check. This, however, would put enormous strain on the peer-review process, and, frankly, the point of peer review is to evaluate the science, not to detect fraud. This science is taken as being true, and this trust is essential to science. Furthermore, manuscripts that have a large volume of results have a correspondingly large volume of supporting data. Thus, it would be unreasonably lengthening the peer-review process. Such data could be made available in the supplemental information, however, so that all readers could perform this check it they desire.