Long-term trends in the Twitter activity.

An OLS regression was used to predict tweet rate using a linear term representing the increasing time of the debate, and a quadratic term over the same time frame, which reflects an increase and then decrease. Both are highly significant, with the quadratic term in general having the larger effect size. That last column shows that the decay portion of the quadratic term still significantly predicts tweet rate when included alone. There is thus a longer timescale process of height activity then decay.Long-term trends in the Twitter activity.