in the past I've been fairly casual about citing citations from other peoples work ... e.g ( [x] cited in [x], 2010) but now that I'm starting to write papers which I hope to have published I've become perhaps a little more attuned to the fact that this doesn't seem to be common practice. But then you look at the reference lists for some of these papers and cant help but feel there is no way this person has read all this material.

So the question is, within the social sciences, what is your view of citing books/articles that you haven't read but wish to quote from (based on reading that quote in someone else's work)? What is accepted if not spoken practice? Or is it preferable to stick to writing "cited in" ?

In an ideal world we go back to each original source, cite and reference as appropriate.  Where it is not feasible to access the original source the text with the citation of the original article can be cited by referencing as 'cited in Smith 2010...' This can be a dubious way of referencing as the later author could have misinterpreted the original article, so caution should be taken when doing this. Most publications would prefer to see original sources cited as this also indicates the quality of the authorship.

It's not really considered good practice to do it very often. One or two in a paper would probably be okay but any more would not be good. In the main you are expected to go back to the original work, certainly for your thesis. If citing somebody citing somebody else is part of the general preamble etc it's not so bad but if it's part of your core argument, then no!

Of course, I have read every word of every sentence of every work cited in my 20 page PhD bibliography - not! It's one thing I am really worried about come the viva!

On a similar note (hope you don't mind me putting it here) I've been concerned about citations where I've got the information I need form the abstract but have not been able to get the whole article. I've cited it as if I had got the whole article where I haven't been able to.

As for the OP I've tried to go for the original source, though have gone for secondary sources where I have read review articles, as otherwise I'd have hundreds on the same topic! I've found amazon quite good for getting to look inside some books where I've wanted to cited them but my library hasn't had the original. Although it doesn't quite count as reading thw hole book it's given me enough to feel I can cited it as a primary source! Otherwise if there's just a few important ones you could go for an interlibrary loan to get the book.

I never secondary-cite, always use primary sources. For my entire PhD I have only read about 2 articles cover to cover, the rest i just dip in and out of, or just read the abstract. I also use google books to read a sentence - i will then cite as a primary reference. I don't think anyone ever said you had to read the whole article/book in order to cite them. I think as long as you understand what they're saying then that's fine :-)

Realistically citations are there so someone can read the work and say "ah yes, I will go and have a look at that work, where she is saying she read that" so as long as its faithful to that i.e. you haven't made up details and pretended they were in an article when they weren't then that's fine.