An important point about this result: the significance is high because they had a huge population to test on, but the effect size is very small. The effect size is 0.001 which means that if one of your friends posts something negative (or positive), you have a 1/1000 chance of also posting something negative (or positive) as a result that you wouldn't otherwise have posted.
Anyone want to do a power calculation for an effect size of 0.001 and a sample size of 700,000?
Eyeballing it, I’m not convinced that’s enough.