Ben Goldacre of Bad Science recently asked "why don't journalists link to primary sources?". Science journalists often - but not always - write an article based on a published, peer reviewed scientific journal article.
Sometimes not, though. Sometimes an article is a feature - based on a number of sources. Sometimes it's based on a press release on some pilot or unpublished work, or perhaps a report submitted when a research grant ends, plus an interview with the scientisist involved and (one would hope) another couple of experts. Sometimes it's based on a conference presentation or press conference or similar.
Having seen this from both sides I'd say there are a number of reasons. Some are administrative or technical, some definitely journalistic, some due to laziness, and some done with the best of intentions.
In some cases, the press release was only made available via email. It might be possible to put it up on the newspaper's/journalist's website. I may be misrepresenting this situation - it's possible that all the press releases that come round by email also end up on some website or other - but I'm not sure. Let's say that at least some press releases are not ultimately stored on the web. Certainly, before the embargo deadline, they aren't publicly available.
This leads on to a major category of administrative reasons why journalists don't link to papers and press releases. They have been told not to yet, and they don't have the link yet. If you are writing an article for an embargo of 8pm and you have written it, proofed it, and your editor has signed off on it, and someone else is going to put it up on the website for you because you have to go home/work on something else, it's really unlikely they are going to search out the (non-live till that exact moment, and possibly later) link to the brand new article. Sure, you can edit a web article the next time you are able, but where you've already put in a link to the journal or scientist's website, or both, and you are busy with other stuff, sometimes this might not happen. Plus, there is the whole paywall issue.
Of course, if the article is based on something else such as a presentation or several interviews, there's a chance these aren't available at all - not everyone is happy with their presentations being recorded, not for sinister reasons - some inexperienced presenters are simply shy. There's a big difference between 30 academics hearing your every word and a recording being made for posterity. The slides you write for a presentation might make sense to the audience, but they could be in gibberish for all other purposes.
The second category is the lazy one. Regurgitation of press releases without any further research. I'm guessing errors like assuming "heels" means "shoe heels" fall into this category.
I would imagine a journalist who rehashed a press release without talking to the scientists or reading any articles would not really want this known by all and sundry - especially if the press release is barely re-written. Not all press releases are perfectly transparent (shock horror!) and sometimes this is the fault of a scientist, sometimes a press officer, sometimes just one of those things where no-one involved in writing something considers what alternative meaning it could have.
I do think it's possible to write something informative for the general public based on publicly available materials without conducting any additional interviews. Otherwise, everyone would just go to the primary sources, and there would be no point in journalism. Leaving aside the point that the general public doesn't have the resources to know everything about what is out there happening at any given instant (even in science), it seems pretty unlikely that any one person - even with scientific training - could understand the full implications of scientific articles in every single field. That, as I say, is why we have journalists.
Perhaps if the scientist in question communicated with the reader and explained the article, they could understand it - but given the number of people who contact me with weird queries that make no sense after I either write a popular science article or my own work is reported on, I'm just hoping that there are at least some people for whom the article made sense on its own and who therefore didn't contact me, and thus saved me a bit of time.
I think this point is linked to my third (and you'll be happy to hear, final) reason why journalists don't link to primary sources. They really genuinely think there is no point. Sure, the geeky readers of Bad Science want to read some primary source articles. They might represent, what, 0.1% of the newspaper reading public (that is a random figure plucked from my head, before you ask). But I bet those readers don't bother looking at the primary source article more than, say 1% of the time. And if they do (and I have seen this in the Bad Science comments section, where an article has been in my general field) they often show that they think the article was badly designed/run/reported when in fact they are just displaying they know little about experimental design/conduct/analysis. No offence, Bad Science readers, and of course this is very much not a reason for non-specialists not to read original articles - but though sometimes you have spotted something the author missed, or the journalist tried to sweep under the carpet - sometimes you haven't.
I have noticed that many science bloggers rant about this whole issue, and few science journalists (that's not to say no science journalists, but few). Privately science journalists I've spoken to feel more or less as I have summarised. One thing to note particularly about science bloggers is that some of them - by no means all - don't write in a way that would make sense to a non-specialist. I love some of the debates on specialist science blogs, but frankly, you could not possibly pretend they are for the non-specialist reader. General readers prefer newspaper and BBC reporting because it's aimed at them. That's who the (non-ranty) parts of my blog are aimed at too (and I will write some more evidence-based parenting articles some day, honest).