Why cheating in science is so bad

I have been hammering away at cases of cheating in science, trying to make them more widely known. I feel that this is a really important issue for a crucial reason.

One of the great strengths of science is that it is global. Scientists all over the world contribute to its knowledge base, enabled by their shared paradigms that are the hallmark of science. Another strength is the trust that they have in each other’s work. Scientists assume that the work of others has been honestly done and reported and can be relied upon. They rarely replicate the work of others, unless it is to understand it better, as a test of their own methods and apparatus, or it is so surprising that it arouses skepticism. It is this trust that enables science to move fast.

Peer reviewers do not actually replicate the work but check to see if the work follows accepted protocols, addressed all the relevant issues, took into account and acknowledged the work of others, and the conclusions were justified by the data on the basis of theory and reason. They do not challenge the new data unless there is a serious problem.

This is what makes cheating in science so tempting to some and why anyone who does so is committing a very serious offense. The errors in their papers can usually be detected and exposed, especially if the works deals with important questions. Much harder to repair is the sense of trust. If that goes, science will be slowed down considerably because of the need to always check other people’s work.


  1. Enkidum says

    I’ve been reading all your posts on cheating recently and very much enjoying them (well, “enjoying” isn’t quite the right word, I mean they piss me off a lot, but I’m glad you wrote them).

    I’m a cognitive psychologist, and there’s been some cheating going on in our field – there was an especially egregious case last year in social psychology: http://chronicle.com/article/As-Dutch-Research-Scandal/129746/.

    But I have to say I was a bit surprised to hear you say that scientists don’t generally replicate the work of others. That’s definitely true in psychology, but I thought we were outliers there. I was under the impression that physics results, for example, weren’t really taken seriously until they’d been independently verified. But I guess that’s not true?

    Either way, I’d say lack of replication is a huge problem – I think we really should be devoting more of our time to replication. Because ultimately, that’s the only thing that allows us to know what’s true. Statistics, smart research design, clever arguments – none of these mean a damn thing if the results can’t be replicated. And especially in a field like psych, it turns out that replication of many results is really, really hard.

  2. Psychopomp Gecko says

    It also makes science look terrible whenever someone takes the easy way. Believers will always try and talk about “what about this fabrication, what about that fabrication” but we can point out how science doesn’t accept false evidence and that scientists look out for that kind of thing. Except when they don’t. Then the process looks a lot less reliable.

  3. VincentFleury says

    As a physicist I agree with what is said above especially :

    “I’d say lack of replication is a huge problem – I think we really should be devoting more of our time to replication”

    it seems to me there is a slow change in science; replication is no longer an issue at all;
    On the one hand, it would be impossible to find any funding in order to replicate an experiment. All grants require “original” work, “ground breaking data” “new frontiers ” etc. so there is no way one can fund replication.
    In another hand, it seems to me all scientific litterature is pretending more and more to be “perfect”, and without any error; the articles are more and more written in a way which suggests that what is found is an absolute truth, obtained often with gigantic technical means. Therefore, there is again no point in replicating that.

    In the end, contemporary science is at risk of building up giants with clay feet, which to some extent is happening in many disciplines.

    Also in biology you find this bizarre tendency : you ask someone a detail about a paper, more often than not he or she will say, “oh that was not reproducible, we tried many times, and dropped the thing, it was an artefact of staining etc.”, but the original article is never corrected.

  4. Ysanne says

    Even in maths, where checking a paper for sound reasoning is essentially the same as replication, there are pretty fundamental papers around with bugs that just never officially get repaired… usually something along the line of “ok but how do I know this cut-off function is really in the admissible class of functions” that a nitpicky grad student would pick up while trying to understand the technique in depth. Mostly there are is a more or less known fix, for some it’s just a shrug and “yep, I found that one too, but has an unpublished work-around, ask them” from one’s supervisor, and for a tiny few it’s “yeah, that often-cited theorem? actually it’s wrong”.
    This non-existent culture of correcting errors is scary, and doesn’t exactly further the reputation of peer review.
    And it’s getting worse, with crap journals whose names sound similar to good ones getting founded by the dozen, and publishing papers that range from plagiarism and simple wrongness to complete gibberish, as long as the author pays a fee.

  5. NitricAcid says

    I remember spending a great deal of my graduate studies trying to replicate other people’s work, and finding that I couldn’t do it. Once was a grad student who had misinterpreted his observations, and described the reaction poorly. (He said he reacted A and B to make C, which he then reacted with acid to make D, and there were a number of funky colour changes (orange->green->red->yellow) between C and D. I reacted C with acid, and got not D, but E, which rearranged to form G and then X, with no funky colour changes (yellow->yellow). What he had done was mix A and B (which makes a small amount of green B.5 initially, and only slowly forms C), and then immediately add acid (which reacted with the unreacted A to give a known red compound, which then slowly reacted with B to give D.)

    Another time I was trying to make a compound that had been described by another researcher in two publications. The first was a communication- a short paper that said “Compound A was made by reacting B with excess Grignard reagent.” The second said that the compound was made as described in reference 1 (the communication). That was a waste of a semester.

Leave a Reply

Your email address will not be published. Required fields are marked *