About 32M people in the U.S. tweeted about TV shows last year. And a study out this morning from Nielsen and SocialGuide says that programmers should pay attention: The seemingly random messages Twitter users generate can provide a statistically meaningful early warning about whether a show is catching on. “We expected to see a correlation between Twitter and TV ratings, but this study quantifies the strength of that relationship,” says Andrew Somosi, CEO of SocialGuide, which is co-owned by Nielsen and McKinsey and monitors the links between social media and TV. For example, the firms saw a pattern for shows that had been on at least one season. Ratings for the premiere episode vs the previous season tended to be 1% higher among 18- to 34-year-olds when their tweets about the show were up 8.5%. Similarly, premiere ratings for 35- to 49-year-olds rose 1% when their tweets were +14%. The pattern held for midseason shows: This time ratings among the younger group rose 1% when their tweets increased 4.2% while the older group showed a bump when their messages were +8.4%. “While our study doesn’t prove causality, the correlation we uncovered is significant and we will continue our research to deepen the industry’s understanding of this relationship,” Nielsen EVP Mike Hess says.