What are we doing when we write a scientific paper? To cut a long story short, we are taking part in a conversation between scientists in which, ideally, every participant is seeking a better understanding of the subject at hand.
How do we know when we have achieved a better understanding? Can we prove it? It turns out that we can't. Instead, the best a scientist can hope do is to persuade the people that matter that their new approach does indeed lead to a better understanding than existing approaches. See Wikipedia on Thomas Kuhn for more explanation and references.
Note that this applies to all science, not just social science. See the famous paper of Watson and Crick (1953) for an example of authors clearly engaging in a conversation and seeking to persuade their readers of the fruitfulness of their approach. But note that every good paper does the same.
The above analysis has far-reaching consequences for how we approach scientific writing. When writing, we need to: indicate clearly which conversation we want to take part in; give our take on the current state of knowledge with regard to the question at issue in that conversation; give our take on relevant gaps in knowledge; explain how our approach can help to fill those gaps. Then when we actually get down to work and go into our analysis we must have a sceptical attitude to knowledge. Since the whole scientific process is about improving our understanding of the world, we must not take it as given that existing published work is flawless and thus can be treated as gospel truth. When citing published work to support our own argument we should be extremely clear about what that support is; the idea that X said it's true and therefore it must be true cuts absolutely no ice and is an invalid argument. Never use it!
The 'conversation' metaphor also leads us to conclusions about the structure of a scientific paper. Since the conversation is carried on via dozens, hundreds, or even thousands of voices straining to be heard, it is crucial to attract and hold the attention of the right readers from the start. Furthermore, it is crucial to state right at the start exactly which question is in focus, what our hypothesis is regarding that question, and what our key conclusion or argument is. This helps the reader make a quick decision about whether to actually read the paper. And it also helps the reader enormously -- when working through the analysis -- if she knows in advance where the analysis is headed. Compare to reading a detective story littered with subtle clues about the identity of the murderer, all of which are missed by the typical reader. Only at the end, when all is revealed, does the reader have a chance of picking up these clues, but to do so she would have to reread the book. If the author had started by naming the murderer, many things in the story would have been easier to understand.
For more on scientific writing in the social sciences see Chapter 2 of Good essay writing by Redman och Maples.
A gobbet is a miniature scientific paper. It should therefore have a title, and then begin with a clearly stated hypothesis. It should then analyse this hypothesis using evidence and logical argument. Crucially, it should be written as a contribution to a scientific debate or conversation, and NOT as something to be read by an examiner. It should NOT be a literature review, but should include original arguments. However, the hypothesis and argument put forward must be positioned in the literature, however briefly.