Table 5.

Overlapping data among datasets in Group 1.

Scale	Twitter search	Twitter filtered stream	Twitter sampled stream	Twitter search vs. Twitter filtered stream	Twitter search vs. Twitter sampled stream	Twitter filtered stream vs. Twitter sampled stream	Overlapping across the three datasets
Comparison of all tweets of each dataset
Minutea	52.3% (45.1%, 59.6%)	48.2% (44.7%, 51.6%)	0.8% (0.7%, 0.8%)	28.6% (23.4%, 33.3%)	0.4% (0.3%, 0.5%)	0.6% (0.6%, 0.7%)	0.4% (0.3%,0.4%)
Minuteb	72.7% (49.7%, 95.7%)	64.5% (31.8%, 97.2%)	1.6% (1.3%, 1.9%)	37.7% (23.1%, 52.4%)	0.6% (0.2%, 1.0%)	0.6% (0.5%, 0.8%)	0.6% (0.1%, 1.0%)
Hourb	72.9% (51.2%, 94.6%)	54.7% (17.8, 91.6%)	1.8% (0.9%, 2.7%)	28.4% (9.0%, 47.8%)	0.5% (0.3%, 0.7%)	0.6% (0.4%, 0.7%)	0.4% (0.1%, 0.7%)
Dayb	70.5% (60.1%, 80.9%)	53.4% (28.2%, 78.6%)	1.9% (1.0%, 2.9%)	25.6% (8.4%, 42.8%)	0.5% (0.3%, 0.7%)	0.7% (0.5%, 0.7%)	0.4% (0.1%, 0.7%)
Weekb	70.4% (65.6%, 75.2%)	54.0% (37.6, 70.4%)	1.9% (1.3%, 2.5%)	27.5% (16%, 39%)	0.5% (0.2%, 0.7%)	0.6% (0.4%, 0.7%)	0.4% (0.2%, 0.5%)
Comparison of the tweets with the top 10 keywords of each dataset
Minutea	57.1% (53.4%, 60.9%)	52.1% (44.6% 59.5%)	1.0% (0.9%, 1.0%)	31.9% (26.3%, 37.6%)	0.5% (0.4%, 0.6%)	0.9% (0.9%, 1.0%)	0.5% (0.4%, 0.6%)
Minuteb	74.4% (51.7%, 97.1%)	69.4% (42.7%, 96.1%)	1.2% (0.6%, 2.4%)	44.1% (3.1%, 85.4%)	0.6% (0, 1.0%)	1.1% (0.5%, 1.8%)	0.5% (0, 0.1%)
Hourb	77.4% (72.5%, 82.2%)	52.9% (47.9%, 58.0%)	1.3% (1.2%, 1.5%)	30.3% (26.4%, 34.3%)	0.5% (0.4%, 0.5%)	0.8% (0.8%, 0.9%)	0.3% (0.3%, 0.4%)
Dayb	68.8% (42.5%, 95.1%)	58.4% (49.6%, 67.2%)	1.4% (0.7%, 2.1%)	27.7% (15.3%, 40.1%)	0.5% (0.3%, 0.7%)	0.8% (0.5%, 1.6%)	0.3% (0, 0.5%)
Weekb	71.8% (57.0%, 86.6%)	57.5% (50.1%, 64.9%)	1.3% (0.7%, 1.9%)	28.6%s (17.6%, 36.8%)	0.5% (0.4%, 0.6%)	0.7% (0.6%, 0.9%)	0.3% (0.2%, 0.4%)

^aThe denominator is the “Twitter full archive” dataset;

^bThe denominator is the combination of “Twitter search”, “Twitter filtered stream”, and “Twitter sampled stream” datasets.