Someone has falsified their data, or their survey findings, to claim a result that is not really supported by the evidence. Quite properly, we are shocked when such stories crop up. They clearly do not represent the ethical, or honourable way to do research. On the other hand, given the pressures that exist in universities and research centres to get results and to deliver highly rated publications, it's easy to see why and how these things can happen. That's not at all to defend unethical research, of course, since that doesn't contribute to the generation of new knowledge and understanding of the world - which is what good research is surely all about.
But it does mean that institutions that deliberately put a great deal of pressure on their researchers to 'perform' and 'deliver' must always be on their guard to watch out for this sort of distortion. Moreover, undue pressure on researchers can sometimes lead not just to deliberate falsification, but also to honest mistakes committed in the rush to deliver results.
Such concerns about the quality and reliability of research are probably among the factors behind the increasing tendency for universities to insist that all research must be subject to some sort of ethical oversight, and for funding bodies to require a section on research ethics in their funding application forms. However, most of the new research ethics 'machinery' is actually not so much concerned with the aspect of research referred to above, namely its basic honesty and integrity, since this is still largely taken for granted. And anyway, how could the honesty of research be checked? I suppose the system still relies on peer pressure and the replication of experiments to highlight areas where some falsification or other distortion has occurred, though this is inevitably rather hit-and-miss. It also relies on occasional heavy penalties, like the loss of one's job, to provide negative incentives!
More mundanely, though still importantly, most of the new ethical machinery, at least in the social sciences, is more designed to ensure that where research might involve human subjects, the research is done in a way that involves voluntary and informed participation, without any pressure being placed on potential 'subjects' to take part. Within my own School, for instance, the ethical guidelines for research were recently updated, and it's clear that how we do research involving human subjects was a paramount concern.
This makes good sense, though it did seem to me that we were running a risk of being overly heavy handed in our approach. For the guidelines state that all the research we do must be subject to official ethical oversight, not only that where human subjects are or might be involved. This means that our ethics committee should, in principle, get to know about every single bit of research that goes on in the School. I would be surprised if the system works that well in practice, for I suspect that through a mix of ignorance and common sense, many of us will just get on with our perfectly respectable research without worrying about the ethical dimension - if we make a judgement at all, it will be that the research does not give rise to ethical issues. Now, I know that according to the rules we are not supposed to be the judges in our own cases, but I bet we often are.
For instance, a recent paper I wrote on North Korea (now published) was based on diverse secondary sources plus some of my earlier work on institutions related to transition to a market-type economy in Central and Eastern Europe. I didn't perform a survey, nor did I interview anyone, so although I didn't seek ethical approval for the work (and to be truthful, the idea never crossed my mind) it's hard to see what would have been done differently had I done so. And a good deal of research is like this.
Hence bringing ethical considerations into our research can be highly desirable if it helps to make us more careful and honest in what we do, and if it promotes good practice (as with the treatment of research involving human subjects), but we probably shouldn't push it too far. Nor should we make it overly bureaucratic.