If you've ever viewed Twitter as a gauge of public opinion, a weathervane marking the mood of the masses, you are very much mistaken.
That is the rather surprising finding of a new US study, which suggests the microblog zeitgeist differs markedly from mainstream public opinion.
"Twitter users are not representative of the public," Washington DC think tank, Pew Research Center, concluded.
Experts in Australia, where Twitter comment is regularly used in media reaction to major new stories or a method of interaction for television programs, agreed with the US findings.
"While Twitter can give you a good idea of the extremes of how people feel about certain topics, when it comes to measuring opinion of the general public about major issues, it's pretty useless," Laura Demasi, of marketing firm IPSOS Australia, told AAP.
Pew Research's study examined eight major US news events, including November's presidential election, and compared views expressed on Twitter with national polling.
The two didn't match.
"At times the Twitter conversation is more liberal than the survey responses, while at other times it is more conservative," the study said.
The study highlighted a decision made in California's Federal Court which ruled that laws barring same-sex marriage were unconstitutional.
Almost half of the Twitter conversations about the verdict were positive, eight per cent were negative and 46 per cent were neutral.
But wider public opinion on the decision was more mixed - with 33 per cent saying it was a positive ruling, 44 per cent negative and 15 per cent neutral.
The reason, Pew Research Center says, is that only a "narrow sliver" of the population use Twitter.
A recent study by French social media analysts Semiocast showed there were 140 million Twitter accounts in the US - more than one third of the population.
But users tend to be younger and lean more toward the political left than right, the study said.
Ms Demasi added: "Twitter penetration in Australia is not that big so while at times it seems like the whole country is talking about something, it's really just 50 people and a few hundred or thousand who are listening in."
Dr John Lenarcic, from Melbourne's RMIT University, suggested Facebook may offer a more accurate view of public sentiment.