**3. Results**

All the variables of interest were normally distributed. Mean (standard deviation) values of each ankle biomechanical variable and the stability index for each collar type, which were estimated intra-subject first and then inter-subject, are shown in Tables 1 and 2, respectively.

#### *3.1. Anterior Single-Leg Jump Landing*

The result of the ANOVA indicated a significant shoe e ffect on dorsiflexion ROM (F2,28 = 3.829, *p* = 0.035, Wilks's Λ = 0.675, ηp<sup>2</sup> = 0.639), total ROM in the sagittal plane (F2,28 = 7.554, *p* = 0.006, Wilks's Λ = 0.590, ηp<sup>2</sup> = 0.854), ankle joint sti ffness (F2,28 = 7.431, *p* = 0.009, Wilks's Λ = 0.445, ηp<sup>2</sup> = 0.810), and MLSI (F2,28 = 7.418, *p* = 0.004, Wilks's Λ = 0.382, ηp<sup>2</sup> = 0.884). Post hoc pairwise tests indicated that the high collar resulted in a significantly smaller dorsiflexion ROM, compared to the elastic collar (*p* = 0.031, dz = 0.511) and low collar (*p* = 0.043, dz = 0.446) (Table 1), while a significantly smaller total ROM was observed for the high collar, compared to the low collar (*p* = 0.023, dz = 0.756) in the sagittal plane (Figure 3). The ankle joint sti ffness was significantly larger for the high collar, compared to the low collar (*p* = 0.030, dz = 1.040) and elastic collar (*p* = 0.003, dz = 0.629) (Figure 4). MLSI was significantly smaller for the shoe with the high collar, compared to the low collar (*p* = 0.004, dz = 1.232) (Table 2). No other main e ffects of shoe conditions were detected (Tables 1 and 2).

**Figure 3.** Range of motion (ROM) in the sagittal (**a**) and frontal (**b**) planes for both anterior and lateral jump landings in three shoe conditions: high collar, elastic collar, and low collar. \* indicates a significant pairwise difference between the high collar and low collar; # indicates a significant pairwise difference between the high collar and elastic collar.

**Figure 4.** Ankle joint stiffness for both anterior and lateral jump landings in three shoe conditions: high collar, elastic collar, and low collar. \* indicates a significant pairwise difference between the high collar and low collar; # indicates a significant pairwise difference between the high collar and elastic collar.

#### *3.2. Lateral Single-Leg Jump Landing*

There were significant differences in inversion ROM (F2,28 = 4.344, *p* = 0.029, Wilks's Λ = 0.690, ηp<sup>2</sup> = 0.658), total ROM in both sagittal (F2,28 = 6.404, *p* = 0.009, Wilks's Λ = 0.373, ηp<sup>2</sup> = 0.813) and frontal (F2,28 = 6.655, *p* = 0.006, Wilks's Λ = 0.571, ηp<sup>2</sup> = 0.846) planes, ankle joint stiffness (F2,28 = 3.783, *p* = 0.040, Wilks's Λ = 0.703, ηp<sup>2</sup> = 0.610), and MLSI (F2,28 = 7.554, *p* = 0.041, Wilks's Λ = 0.664, ηp<sup>2</sup> = 0.601) between shoe conditions. Post hoc pairwise tests indicated that inversion ROM was significantly smaller for the high collar, compared to the elastic collar (*p* = 0.028, dz = 0.615) shoe (Table 1). The high collar resulted in a significantly smaller total ROM, compared to the low collar (*p* = 0.001, dz = 0.634) in the sagittal plane (Figure 3), while the elastic collar resulted in a significantly larger ROM, compared to the high collar (*p* = 0.019, dz = 0.873) in the frontal plane (Figure 3). No other pairwise differences were observed for ankle joint stiffness and MLSI (Tables 1 and 2).

## **4. Discussion**

In the present study, we determined the effects of football shoes with different collar conditions on dynamic stability and ankle biomechanical characteristics during anterior and lateral single-leg jump

landings. Our results indicate that the high collar football shoe resulted in smaller dorsiflexion ROM and total ROM in the sagittal plane during the anterior single-leg jump landing, while it also decreased inversion ROM and total ROM in the sagittal and frontal planes during the lateral single-leg jump landing. We also found that ankle joint sti ffness was significantly larger for the high collar football shoe during anterior and lateral single-leg jump landings, which contradicted our original hypothesis. For dynamic stability, only MLSI showed significant di fferences during both landing tasks, which was greater when wearing the high collar football shoe and lesser in other conditions; this is partly consistent with our original hypothesis.

The ankle ROM during the anterior single-leg jump landing suggested that the high collar significantly constrained ankle movement, compared to the elastic and low collars. These findings are consistent with Yang et al. and Rowson et al., who reported that peak ankle dorsiflexion or total ankle ROM during a sagittal maneuver was reduced as collar height increases [13,37]. They suggested that collar height and material play an important role in influencing the flexibility and deformation of the whole shoe [13,37]. Additionally, the high collar basketball shoes with strips of plastic that are positioned at the collar's anterior and posterior to the medial and lateral malleoli showed a more restricted ROM of the ankle joint in the sagittal and frontal planes, compared to no plastic condition [16]. It is noteworthy that the elastic collar could not constrain the ankle movement, which might have been due to the low rigidity or high elasticity of the collar material. However, there was no significant change in the frontal plane's ROM. One possible reason is that our healthy participants might have had few inversion-eversion movements during the anterior single-leg jump landings, because our results detected significant di fferences in inversion and total ankle ROM in the frontal plane between the high and elastic collar, but not between the high and low collar during lateral single-leg jump landings. The elastic collar, similar to ankle taping, likely provides a feeling of confidence and stability [18,24]. This result, in our perspective, is in disagreement with a recent report that indicated that high collar basketball shoes do not restrict the peak inversion angle (29.3◦ vs. 28.3◦) and ROM (17.4◦ vs. 15.2◦) in a self-initiated drop landing on an inversion platform [14]. However, our findings are supported by Richard et al., who found that a high collar football shoe e ffectively reduces the amount of inversion by 4.5◦ (38.1◦ vs. 42.6◦) after an inversion platform drop [17]. It is possible that a self-initiated drop landing on an inversion platform does not reach the limitation boundary of the inversion for a high collar basketball shoe. During side-step cutting, Liu et al. and Lam et al. found that the ankle inversion angle, peak inversion velocity, and total inversion ROM are reduced as collar height increases [11,12]. Therefore, there is a restricted angle for an inverted ankle joint position, which might e ffectively increase ankle joint stability and reduce the risk of ankle sprain injury [11,12]. In our study, the dorsiflexion and total sagittal ROM showed moderate-to-large e ffect sizes with the high collar, compared to the other collars. Therefore, the football shoe's higher collar height used in this study could constrain ankle dorsiflexion and the inversion angle during both longitude and widthwise tasks, potentially reducing the risk of ankle sprain injury.

Several prior studies have examined the e ffect of collar conditions on ankle kinetics. Lam et al. detected no di fference from collar conditions on the ankle inversion moment during side-step cutting [12]. In addition, Yang et al. reported that high collar basketball shoes could reduce the plantarflexion moment during lay-up jumps, but not drop jumps [13]. The authors suggested that these di fferential findings were caused by di fferent upper limb positions, movement patterns, and force requirements, as well as the coordination of active and antagonist muscles [13]. These findings are in agreemen<sup>t</sup> with our results showing either no significant change or a small e ffect size in the ankle inversion moment for both tasks; however, di fferent jump maneuvers that are high-frequency and risky during practices or matches still need to be tested. Interestingly, ankle joint sti ffness was significantly increased when wearing the high collar football shoe, compared to the other shoes. Theoretically, ankle joint sti ffness is calculated using the change in joint moment divided by the change in joint angle [35]. Although the change in ankle moment was not measured in our study, it is possible that the enhanced ankle joint sti ffness from the high collar football shoe may be due to a decrease in total ankle

ROM in the sagittal plane. Given the primary role that joint stiffness plays in lower limb injuries [38], overuse injuries at the ankle joint might increase as collar height increases.

Our findings also sugges<sup>t</sup> that MLSI is improved as the height of the football shoe collar increases. A couple of studies have examined the effect of collar height on static or dynamic postural stability [18,39]. However, according to previous research, adequate dorsiflexion ROM is essential for dissipating the ground reaction force [40] and has a positive influence on DPSI [30]; these findings conflict with the results of our study. However, evidence from ankle taping and bracing indicate an increased sense of confidence and stability [24]. Inconsistent findings across studies regarding the dynamic stability of ankle taping or bracing might be due to subjects with or without injury [25–28]. Furthermore, although the current study showed a significant difference in MLSI between shoe conditions during lateral single-leg jump landings, post hoc analysis indicated no pairwise difference, and small effect size. Therefore, this phenomenon still needs to be confirmed, and additional quantitative studies on DPSI are warranted.

There are some limitations to the present study. First, only healthy male college football players were recruited as subjects. Players with functional ankle instability may have different responses to shoe collar conditions, especially for DPSI. Second, it should be noted that our current findings were limited to anterior and lateral single-leg jump landings. Future studies should investigate other typical movements that have high injury risk, such as side-step cutting. Third, different types of shoes may have different mass, which could affect biomechanical responses. A better-controlled experiment is to match the shoe mass across conditions. Fourth, the long-term effect of shoe collar conditions on the incidence of lower limb injuries has ye<sup>t</sup> to be examined. Long-term prospective studies are needed. Finally, the current study only focused on the biomechanical changes at the ankle joint, while knee and hip joint kinematics and kinetics and muscle activity data were not collected.
