"Numbers can lie" discusses studies by John Ioannidis that suggest that more then half of all epidemiological studies may be wrong, in that they are later refuted by randomized clinical trials. In contrast, randomized clinical trials appear to be refuted less often. The story explains the differences between these study types and their limitations.
"Scientists debate fixes – and if they're needed" describes suggestions to improve the quality and reporting of epidemiological studies, such as improving the statistical procedures, sharing data so others can double check the study, or better reporting of the results and limitations of studies when they are published.
"Study guide for research" lists the criteria to look for in a medical study to determine how solid it is. Examples are effect size, statistical significance and study size.
See the archived story and contest questionnaire about how the story was written.