Ellynne:
>Subscribe to several academic, on-line magazine services
>(ones that index actual, real, flesh and blood [figuratively
>speaking] periodicals). Type in various key word phrases,
>doing either a text or subject search, cross reference the
>material under the subject headings that appear for each article,
>and (if so inclined) double check both authors and the periodical
>for overall reliability (Itend to get a bit sloppy on this last, but
>[credible] peer review is an important part of the evaluation
>process).
When doing meta-analyses or similar (ie attempting to pull together the
results of multiple independent studies) four things appear to be
essential:
(0) to formulate the precise question or questions *in advance*;
(1) to decide on tightly defined criteria for study inclusion and exclusion
*in advance*;
(2) to collect ALL studies that fall within the criteria (including those
not published in English) by several different exhaustive search techniques
and find their original source publication (and preferably get hold of the
raw data) -- this usually takes years and involves multiple online and hand
searches, contacting key researchers in multiple countries &c&c&c;
(3) to interpret the results with respect to the question without
preconceptions using some statistically reputable technique.
And even so, the results are usually difficult to interpret. I've read
meta-analyses by different groups of very similar questions with very
similar inclusion/exclusion criteria which came to opposite conclusions.
Tavia