Table 2.
Back-transformed least square means from the multivariable modeling of rein tension (N) in walk (data from 6 riders and 18 horses, n = 21,008 data points/208 normalised strides)
| Variable | Category | LS mean | Group P value | Significant within-category | ||
|---|---|---|---|---|---|---|
| Rein | Left | 12.5 | 0.67 | |||
| Right | 12.8 | |||||
| Nose direction | In | 12.3 | <0.0001 | c | ||
| Out | 12.9 | c | ||||
| Nose angle ROM (degreesa) | <12 | 14.1 | 0.01 | a | b | |
| ≥12–16 | 13.8 | a | b | |||
| ≥16–20 | 11.8 | |||||
| ≥20–24 | 12.9 | |||||
| ≥24 | 10.6 | b | b | |||
The model also contained the fixed effects of stride percentage (P < 0.0001) and its interaction with rein (P < 0.0001). If pair-wise comparisons within a variable were associated with P < 0.0001 this is marked with ‘c’, if 0.01 < P ≥ 0.0001 then ‘b’ and if P < 0.01 then ‘a’, these letters are indicated in both categories
a The nose angle categories contained from top to bottom: 4747; 3939; 6666; 3434 and 2222 observations. Rein tension was modelled square root transformed, hence confidence intervals could not be produced on the back-transformed scale