Exploring Links between Online Activism and Real-World Events: A Case Study of the #FeesMustFall

Table 5

Changes in the average sentiment.

Date	Confidence interval (dd-mm-yy)	From	To	Confidence level (%)	Level
23-12-2015	(29-10-15, 31-12-15)	0.027693	0.077219	98	7
04-01-2016	(28-12-15, 11-01-16)	0.077219	0.012388	98	6
27-01-2016	(19-01-16, 11-03-16)	0.012388	−0.04478	100	9
20-03-2016	(28-01-16, 21-04-16)	−0.04478	−0.0052	99	5
24-04-2016	(25-03-16, 08-05-16)	−0.0052	0.060491	98	7
11-05-2016	(05-05-16, 15-05-16)	0.060491	−0.0997	100	8
23-05-2016	(15-05-16, 31-05-16)	−0.0997	0.014568	97	7
20-09-2016	(24-05-16, 29-09-16)	0.014568	−0.01755	96	8
10-10-2016	(26-09-16, 30-01-17)	−0.01755	−0.04232	100	3
01-02-2017	(28-01-17, 11-02-17)	−0.04232	0.052568	96	4
24-03-2017	(02-02-17, 03-04-17)	0.052568	0.004263	100	5

(Confidence level for candidate changes = 50%, confidence level = 95%, CI = 95%, bootstrap = 10 000, without replacement, MSE estimates).