Perhaps ±2.4 is the estimated population standard deviation of the tweeted group, rather than a confidence interval. The confidence interval could be the smalle

Perhaps ±2.4 is the estimated population standard deviation of the tweeted group, rather than a confidence interval. The confidence interval could be the smaller value you seein the charts. Two distributions can overlap a lot and still have a low p-value for anova etc if there are enough data points.