As most, if not all, of you will know scientific meeting abstracts can make up an important part of scientific research. For those that are unaware, well formal scientific meetings, researchers who are attending send off short abstracts of their work that they will present at the meeting (if you have not seen one before a quick google will give you a good idea). Usually this is just a couple of hundred words describing the very basics of the work, though occasionally you get longer ones with reference lists and even figures. This means that in advance of the meeting and the talk itself you have an idea of what is being discussed which helps understand the presentation itself. If it’s a big meeting where you can’t attend every talk, then the abstract can help you decide which talks you want to see (or avoid) and afterwards you have a record of who said what about what subject. In theroy at least, and that is the problem.
Short abstracts are not peer reviewed and of course they do not come with reference lists, data sets, statistics, detailed methodologies or anything else that you would expect in a formal paper. However, researchers still cite them in actual scientific papers just like they would with normal papers and people will happily discuss them (especially online) as they would any other paper. Of course it is understood that they do not have the same authority as papers, but I would argue that in many cases they simply should not be used at all.
The fundamental lack of information, data and peer review means that you can have no confidence in whatever was said in an abstract. Yes of course I trust my colleagues (well, most of them) to produce good science to the best of their ability, but without a dataset I (or anyone else) can’t check that they didn’t make a mistake somewhere, or excluded an important datapoint, or used the wrong statistics, put the decimal point in the wrong place or whatever. Without peer review, we can’t ensure that there is no deliberate or acidental misleading of the reader theough manipulation of the work, or a balanced view of the evidence or whatever. Often these abstracts are followed up by formal papers which do provide the level of detail and review that is necessary, but of course many never do become papers, some take several years and are left as abstracts, and some researchers often cite abstracts in addition to the actual paper that resulted from them.
Possibly most damning of all is the fact that there is no gaurantee that what was written actually reflects the opinion of the author, which is quite important. Abstracts are soemtimes submitted months in advance of a talk being given and in that time you might realsie that your bone belongs to a different species to the one you thought it did, or came from a different statigraphic level, or was put on upside down, or your stats were wrong or whatever. Of course in your presentation you can easily correct this to your audience, but the *written* and published record is still the abstract, and people who missed your talk, or the meeting, or didn’t take notes, or can’t remember what you said will be going from the words on the page, not the words you spoke.
As such, and hopefully understandably, I have a real problem with people citing abstracts most of the time. Yes there are of course extenuating circumstances for a number of them, (such as avoiding claim jumping on people, something I have recently had to do myself) and people are not going to stop anytime soon. However, I would caution both against their general use and treating them as authorities – even their own authors may not think they are especially accurate. The more extravagant or unusual the claims, the more you need to be careful about treating them as ‘science’ in the way that you would a paper – just becuase someone says they have a mid Triassic dinosaur or statistical proof that pterosaurs can’t fly doesn’t mean that they really do.