Tiré de S. Horton, Web teaching .
source: http://www.dartmouth.edu/~webteach/articles/sites.html
Too often it is hard to determine who authored a Web site, never mind his or her credentials. But with the free-for-all that is the Web — with sites authored by everyone from grade-school children to movie fans to political organizations to extremist groups to Nobel laureates — and with no governing body to ensure quality or accuracy, users need to turn their critical evaluation skills to Web site content.
One of the hardest aspects of evaluating Web sites is the lack of standardization. In print, we can normally find information about the provenance of a document — the author's name, information about the publication, usually something about the author's qualifications. On the Web, nothing is required. There is no standard way to indicate a document's origins, so the information is often omitted. Also, many of the visual cues we get from printed materials are not on the Web. The typography and quality of the printing and paper for a scholarly journal, for example, instills more confidence than that of the daily rag. By contrast, Web documents of the utmost reliability may be poorly designed and executed. Without proven evaluative systems, you will need to devise your own methods for selecting quality Web sites.
Here are some things to look for when evaluating Web sites:
Origins. Where is the site from? Is it from a source you trust, such as the New York Times? Is it from an educational or government organization (that is, sites that end in .edu or .gov, such as www.nasa.gov or www.yale.edu)? Or is it impossible to tell from the page where it comes from and who the author is? If there is no indication on the site of authorship or affiliation but you feel that the materials are credible and useful, you may wish to do a bit of sleuthing to determine its origins, such as searching the domain name registry to find out who owns the site's domain.
Freshness. How recently were the materials published? Well-designed sites include a publication date (the date the materials were first written), as well as a last-updated date. Web pages have a short shelf life. Beware of pages that have been untouched for more than a few months.
Bias. Many informational Web pages are sponsored by organizations that are trying to sell something. For example, you may find a page about migraine headaches on a pharmaceutical company's Web site. This association does not necessarily render the materials useless, but it is wise to regard them with some skepticism. Also be wary of sites from not-for-profit organizations (for example, www.freecannabis.org): because their mission is to promote a point of view, the information they offer may not be objective.
Popularity. One way to get endorsement for a Web site is to see who else likes it. You can use link searching to check a site's popularity by seeing who links to it from their pages. For example, in AltaVista, typing "link:www.dartmouth.edu/~compose" will show the sites that link to Dartmouth College's Composition Center home page. If only a few sites link to the materials, and they are mainly personal pages, you might consider looking for an alternate source. If, however, you get many search results and you recognize some listings (perhaps the English department site at your institution), then you have some assurance that the materials are of high quality.