Facebook fandom is a 74 percent to 82 percent accurate barometer of federal election outcomes, the social media site says in a post-election analysis of the mid-terms.

In the case of House races, 74 percent of the 98 hotly contested races tracked by the Facebook analysis ended up going to the candidate with the most fans. Of those races, five are still too close to call.

In the Senate election, about 82 percent of 34 decided races (the outcome in Alaska remains unknown) also went to the candidate with the most Facebook fans.

Some of the races where the Facebook fandom was an incorrect measure of the outcome were among the most high profile of the midterms, however. For example, Sharron Angle and Christine O'Donnell, Republican candidates for the Senate in Nevada and Delaware, respectively, both vastly outnumber their Democratic opponents in Facebook fandom--and both lost.

Even in races that gained far less national attention, Facebook fandom can be a wildly incorrect barometer. In Virginia's 11^th House District, for example, Republican challenger Keith Fimian has more than double the 3,050 fans of Democratic incumbent Gerry Connolly--yet the two candidates are separated by a mere 925 votes, as of Nov. 3. The 11^th district includes most of Fairfax County, home to many presumably technology-savvy federal information technology contractors.

If Facebook fandom by itself is a not-always correct measure of electoral outcomes, a social media company with a tool that monitors social media conversations across multiple platforms claims it correctly predicted seven of eight races it decided to closely follow. The company, Alterian, says it correctly predicted that O'Donnell would fail but made an incorrect prediction in the Texas gubernatorial race, where Republican incumbent Rick Perry beat off Democratic challenger Bill White.

However, in a reminder to not always believe what you see online, a Google app reportedly provided the incorrect polling station for as many as 727,000 households in 12 states included in an analysis by a firm that's not Google that provides its own polling place locator tool.

For more:
- go the Facebook analysis of the 2010 midterms

Related Articles:
Candidate Facebook fandom doesn't always correspond to leading in the polls
Q&A: Wayne Moses Burke on Gov 2.0, Open Government and social media
Spires: Social networks are cybersecurity problem