Someone wrote in to the MedStats email discussion group and asked about where to find interesting articles with full free text and sufficient detail that students could calculate some of the statistics on their own. This person had relied on BMJ (formerly the British Medical Journal), which is a source that I have also found useful.
I mentioned that an even better resource is BioMed Central (BMC) at www.biomedcentral.com. This is a group of journals all published under an open access license. That means that not only do you have the full free text available on the web, but you can use this text in any form as long as you acknowledge the source. I regularly quote abstracts from these journals on my web pages and use them in my training classes. You can find several examples in the Practice Exercises at
I also relied heavily on BMC journals for my book. Almost all of my "On Your Own" exercises come from BMC journals.
You need good search skills, and even then you have to filter through a lot of papers which use the statistic you are interested in but don't provide enough detail for you to use it as a teaching example. I was recently looking for some real data with which I could illustrate computation of Cohen's Kappa. It was a long tough slog through dozens of articles before I found a good example (www.biomedcentral.com/1472-6963/6/115), which I plan to use when I next update my page defining kappa and showing an example calculation.
You should also learn how to use the LIMITS tab in PubMed, because this allows you to search only within journals with full free text on the web.
I still use bmj.com a lot, but not as much as recently because they have changed their access policy from where everything was fully available on the web to one where a few things are freely available from the get-go, but others you have to wait until the article has sufficiently aged. It's still probably your best choice if you want to look at just a single journal, because they have so many articles on research methodology. A close second would be the Canadian Medical Association Journal (www.cmaj.ca), which still offers full free text for everything, and which also offers a fair number of articles on research methodology.
You might want to look at some of the nice data set libraries that are on the web. My two favorites are the Data And Story Library (DASL) at lib.stat.cmu.edu/DASL and the Australasian counterpart (OZ DASL) at www.statsci.org/data.
Also worth noting is that the open source statistics textbook project (www.massey.ac.nz/~mbjones/Book) is looking for people to contribute examples and exercises.