Summary of tracking statistics
| Videos tracked | Ant/Post Bias | Deviation from random | Comparison (Watson test) | ||
| Percent Ant (n) | Percent Post (n) | Rayleigh test | |||
| % | % | ||||
| Global EB1 | 42.0 (8,613) | 58.0 (11,881) | 1 × 10−133 | +++ | Global EB1 versus Staufen: 0.001 < P < 0.01 (some similarity) |
| Global Staufen | 44.1 (215) | 55.9 (273) | 3.2 × 10−4 | + | |
| par-1 EB1 | 49.4 (2,811) | 50.6 (2,879) | 2.5 × 10−3 | − | par1 versus global EB1: P < 0.001 (different to global EB1) |
| Global EB1a | 42.0 (8,613) | 58.0 (11,881) | 1 × 10−133 | +++ | |
| Ant EB1 | 46.4 (3,824) | 53.6 (4,416) | 1.7 × 10−10 | ++ | Ant versus mid: P < 0.001 (different) |
| Mid EB1 | 39.6 (2,188) | 60.4 (3,341) | 4.2 × 10−65 | +++ | Mid versus post: P < 0.001 (different) |
| Post EB1 | 36.9 (1,466) | 63.1 (2,506) | 1.4 × 10−79 | +++ | Ant versus post: P < 0.001 (different) |
| Ant EB1a | 46.4 (3,824) | 53.6 (4,416) | 1.7 × 10−10 | ++ | |
| Videos tracked | Ant/Post Bias | Deviation from random | Comparison (Watson test) | ||
| Percent Ant (n) | Percent Post (n) | Rayleigh test | |||
| % | % | ||||
| Global EB1 | 42.0 (8,613) | 58.0 (11,881) | 1 × 10−133 | +++ | Global EB1 versus Staufen: 0.001 < P < 0.01 (some similarity) |
| Global Staufen | 44.1 (215) | 55.9 (273) | 3.2 × 10−4 | + | |
| par-1 EB1 | 49.4 (2,811) | 50.6 (2,879) | 2.5 × 10−3 | − | par1 versus global EB1: P < 0.001 (different to global EB1) |
| Global EB1a | 42.0 (8,613) | 58.0 (11,881) | 1 × 10−133 | +++ | |
| Ant EB1 | 46.4 (3,824) | 53.6 (4,416) | 1.7 × 10−10 | ++ | Ant versus mid: P < 0.001 (different) |
| Mid EB1 | 39.6 (2,188) | 60.4 (3,341) | 4.2 × 10−65 | +++ | Mid versus post: P < 0.001 (different) |
| Post EB1 | 36.9 (1,466) | 63.1 (2,506) | 1.4 × 10−79 | +++ | Ant versus post: P < 0.001 (different) |
| Ant EB1a | 46.4 (3,824) | 53.6 (4,416) | 1.7 × 10−10 | ++ | |
Ant, anterior; Post, posterior. In the Rayleigh test for uniformity of directional data, the degree of deviation from random is given (Mardia and Jupp, 2000). Results of the Rayleigh test are displayed on a scale from random (−) to not at all random (+++). In the Watson two-sample test for homogeneity of two samples of circular data, the higher the p-value, the more similar the populations (Mardia and Jupp, 2000).
Duplicate data entries included to clarify the pairwise comparisons by Watson test.