What's in a good abstract?
by Professor David Dalton
In General
A good abstract contains the same elements of a complete paper - namely an introduction, clear statement of the hypothesis, methods, results, and discussion. Generally, citations are not used. The cardinal rule is: BE CONCISE. Short declarative sentences work best. Lengthy or wordy abstracts should be avoided as they will be much less effective in catching the reader's attention. Every word and phrase should convey some central meaning. Writing an abstract requires a lot of concentration - it's the most difficult part of most manuscripts or theses to write because you must economize your words and go straight to the heart of the matter. No b.s. You can almost always make any abstract better by going back through it and carefully editing out the fluff. Avoid generalities and vagueness. Provide information-dense results.
Consider this example:
Success in academia is hypothesized to require specific phenotypes.1 In order to understand how such unusual traits arise, we used human clones to identify the molecular events that occur during the transition from a graduate student to professor.2 A pool of graduate student clones was subjected to several rounds of random mutagenesis followed by selection on minimal money media in the absence of dental insurance. Students surviving this selection were further screened for the ability to work long hours with vending machine snacks as a sole carbon source; clones satisfying these requirements were dubbed "post-docs". In order to identify assistant professors from amongst the post-docs, this pool was further mutagenized, and screened for the ability to turn esoteric results into a 50-minute seminar. Finally, these assistant professors were evaluated for their potential to become full professors in two ways: first, they were screened for overproduction and surface display of stress proteins such as Hsp70. Assistant professors that displayed such proteins (so-called "stressed-out" mutants) were then fused to the M13 coat protein, displayed on phages, and passed over a friend and family members column to identify those that were incapable of functional interactions.3 These were called full professors. Although these mutants arose independently, they shared striking phenotypes. These included the propensity to talk incessantly about their own research, the inability to accurately judge the time required to complete bench work, and the belief that all their ideas constituted good thesis projects.3,4 The linkage of all these traits suggests that these phenotypes are coordinately regulated.5 Preliminary experiments have identified a putative global regulator. Studies are currently being conducted to determine if overexpression of this gene product in post-docs and grad students can speed up the grad student-full professor evolutionary process.6
- A very clear, concise statement of the hypothesis. Generally this should be preceded by some sort of background information.
- The methods, starting with the general and then going into specifics. This example probably has too much detail on methods, but most of it is necessary to understand the work. Don't go beyond what is necessary for the reader to understand the results.
- A nice example of how to write concisely using parallel construction to get a lot of information in a few words.
- Clear, concise statement of results
- Succinct discussion...doesn't beat around the bush
- In some cases (but not all) it may be appropriate to offer some indication of where the work is going.