Can Tweets drive television ratings?  Meaning, when people are gushing about a show, does it change other people’s choices and behavior?  If it does, it means we know something new about how social media activity affects events in the real world.

You’d think we’d know the answer by now.   More than $70 billion gets spent on U.S. TV advertising every year. Anything that provably drives up viewership a percentage point or two is worth big bucks.  So a whole crop of analytics startups have arisen in recent years to chase these dollars by analyzing what it means when people Tweet about things they see on TV, as they tend to do (see “A Social Media Decoder”). (In fact, the company featured in that story, Bluefin Labs, was bought by Twitter.)

The hard science on Tweets and ratings isn’t quite there.  Nielsen, another major analytics company, today released a study on the question, performed with SocialGuide, a Twitter analytics company now part of NM Incite, a joint venture between Nielsen and McKinsey and Company.  You can read it for details, but the bottom line is this: they can’t say conclusively that Twitter activity actually causes changes in TV ratings.  They do, however, see a statistically significant correlation.

This question may mainly be of interest to the entertainment and advertising industries, but it also speaks to the limits of what we know about how online social activity drives events in the real world.  It surely does in some cases: one dramatic and highly structured study showed that certain Facebook messages on Election Day can drive real-world voting turnout (see “How Facebook Drove Voters to the Polls”).  But in the messy world of Tweets—despite all the money and startup activity devoted to the analytics–the answers are still a little murky.