What neither Twitter nor Nielsen tells you is that "statistically significant" means that the change is bigger than the statistical error rate (meaning it probably remains super-small or Twitter would be giving us the number) and that the change measured is short-term, not long-term.
I doubt there is any real correlation as Siffy showed with Sharknado that you could orchestrate a great Twitter campaign and not get a single more viewer.
To me this is a flawed study that was done to please the networks (who'll use it to renew their failures with a new excuse) and twitter (which will claim it controls the future of TV.
Sad to see Nielsen lower itself to such nonsense.,