Small changes in ball position at address cause a chain effect in golf swing

The purpose of this study was to investigate how the ball position along the mediolateral (M-L) direction of a golfer causes a chain effect in the ground reaction force, body segment and joint angles, and whole-body centre of mass during the golf swing. Twenty professional golfers were asked to complete five straight shots for each 5 different ball positions along M-L: 4.27 cm (ball diameter), 2.14 cm (ball radius), 0 cm (reference position at preferred ball position), – 2.14 cm, and – 4.27 cm, while their ground reaction force and body segment motions were captured. The dependant variables were calculated at 14 swing events from address to impact, and the differences between the ball positions were evaluated using Statistical Parametric Mapping. The left-sided ball positions at address showed a greater weight distribution on the left foot with a more open shoulder angle compared to the reference ball position, whereas the trend was reversed for the right-sided ball positions. These trends disappeared during the backswing and reappeared during the downswing. The whole-body centre of mass was also located towards the target for the left-sided ball positions throughout the golf swing compared to the reference ball position, whereas the trend was reversed for the right-sided ball positions. We have concluded that initial ball position at address can cause a series of chain effects throughout the golf swing.

www.nature.com/scientificreports/ could have detrimental effects on the remainder of the swing. Any compensatory movement or counterbalances in the golf swing were results of poor posture' 19 . These coaches clearly understand that there is a chain effect in golf swing, which is influenced by the initial state at address. Kim et al. 20 studied the effect of the ball position along the mediolateral (M-L) direction, and found that the left-sided ball position causes open shoulder alignment at address as well as changes in club-head movement at impact. Another study by Bradshaw et al. investigated golfers with different handicap levels and found that the variability of the M-L ball position is related to the variability of the club velocity at impact 21 . Similarly, Zhang et al. found that the variability of the M-L ball position is related to the variability of the ball launch angle after impact 22 . Although these studies have revealed the importance of the ball position regarding the initial setup and final outcomes of golf swing, our knowledge on the influence of M-L ball position on the golfer's movements during the course of golf swing is largely limited. A comprehensive study on the ball position and golf swing can potentially offer an opportunity to better understand the influences of the initial state on the consequences of swing behaviour. Furthermore, the outcomes of such a study can provide golf coaches and players with the knowledge of golf swing mechanisms in relation to the ball position and the insights into the improvement of coaching strategies, and eventually, golfer's performance. For example, if a golf ball curves dramatically in flight from left to right for right-handed golfers (called 'slice' in golf), golf coaches and players often attribute it to insufficient shoulder rotation 23,24 , whereas a mal-positioned golf ball may be the cause of a chain effect: an open shoulder alignment 20 leading to insufficient shoulder rotation 19 , and eventually, slice.
Contrary to other previous studies that focused more on the influence of M-L ball position on address position, club-head velocity at impact, and ball trajectory after impact, the current study investigates the behaviour of the golfer during the course of golf swing between address and impact. Our study examines the influence of the M-L ball position on the kinetic and kinematic variables considered to be critical in golf swing mechanism, such as ground reaction force (GRF), centre of pressure (COP), body segment and joint angles, and whole-body centre of mass (COM), during swing 1,25 . We hypothesise that a small change in ball position in the M-L direction at address will cause a chain effect through the course of the golf swing.

Methods
Participants. Twenty professional male golfers (mean ± standard deviation: age; 29.25 ± 4.20 years, height; 1.78 ± 0.06 m, mass; 77.96 ± 10.48 kg) volunteered to participate in this study. The participants had no history of chronic pain, major injuries, or surgery in at least the last 6 months. All procedures were approved and performed in accordance with the ethical standards of the Yonsei University institutional review board (IRB#1040917-201601-SB-104-02). Written informed consent was obtained from the participants before the experiment.
Procedures. Thirty-five reflective markers with a 14-mm diameter were positioned on anatomical landmarks, based on the Vicon Plug-in-Gait full-body model (Oxford Metrics, Oxford, UK). Additionally, four reflective adhesive tapes were attached to the five-iron golf club (1 tape on the toe of the club head, 1 tape on the hosel, and 2 tapes on the shaft). Reflective adhesive tape was used on the golf ball.
The participants completed a self-selected warm-up with stretching and several golf shots for a minimum of 10 min. They performed an initial static, a standing calibration trial. Each participant was then asked to assume their preferred address position, and the ball position was marked invisible to them while the outlines of each foot was visibly drawn on the force platforms for consistent foot positions over multiple swings. The participants were asked to perform a straight shot that they would usually hit on a golf course towards a net 5 m away from them consistently over multiple trials. The participants performed five trials for each of five different ball positions along the M-L direction: 4.27 cm (left from the reference position) (LF), 2.14 cm (LH), 0 cm (reference position), -2.14 cm (RH), and -4.27 cm (RF) (Fig. 1). For each trial, the ball position was changed randomly and blindly. The diameter and radius of the golf ball were chosen as 4.27 cm and 2.14 cm, respectively, to displace the ball from the reference position. Thus, each participant performed 25 trials in total, five trials at each ball position. Additional shots were given to participants when they could not complete the full swing (2% of all shots) to complete five shots for each ball position condition. The laboratory coordinate system was set such that the X was anteroposterior (A-P) axis, Y was M-L axis, and Z was vertical axis (Fig. 1) following the right-hand-thumb rule.
The address was defined as the frame immediately before the club initiates its movement away from the ball. In previous studies, the time when the club-head changes its direction along the mediolateral direction was identified as the transition from backswing to downswing (i.e. TC) 22,27,28 . However, the lower extremities typically start downswing actions before the change in club-head direction. To capture the initiation of downswing by the lower extremities, we added one more event, TP, which was identified as the frame when the pelvis changed its rotational direction. Impact was defined as the frame at which the club makes contact with the ball. The other events (B45, B90, B135, B180, B225, D225, D180, D135, D90, and D45) were identified using the angle of the club   www.nature.com/scientificreports/ found elsewhere 33,34 . We performed SPM on each dependent variable between the address and the impact. The critical significance was set at p < 0.01 instead of a more traditional p < 0.05 to partially compensate for the inflation of statistical error associated with multiple variables as suggested in other previous golf studies 11,27,35 .

Results
SPM analysis showed that the ball position was associated with systematic changes of did not show statistical differences between ball positions (Fig. 3). The results are presented below in the order of GRF variables, body segment and joint angles, and whole-body actions. More detailed mean ± SD for each variable at each event can be found online in the supplementary Table S1-S17.

Discussion
We examined the effects of M-L ball position on GRF, body segment and joint angles, and whole-body actions during golf swing, and we found that the M-L ball position systematically influenced the GRF Z,L , GRF Z,R , GRF Y,R , GRF X,R , COP Y , COP X , A S , A P , A A,L , A A,R , COM Y , and COM X (Fig. 3).
At address. Our findings on the differences between ball positions at address are consistent with a previous study 20 . Kim et al. (2018) showed that the left-sided ball position was associated with a greater/smaller vertical GRF on the left/right foot, and a more open position of shoulder segment. In a quasi-static posture at address, the whole-body position demonstrated by COP and COM in the M-L direction can be predicted from the difference in the vertical GRFs of feet (GRF Z,L and GRF Z,R ). Consistent with the mechanics, our study reports that the whole-body position demonstrated by COP Y and COM Y was towards/distant to the target in the M-L direction (Fig. 4a,b). We also found that the shoulder angular position (A S ) showed systematic differences between the ball position at address. The shoulder in the left-sided ball position was more open than that at the reference ball position (Fig. 4a), whereas the trend was reversed in the right-sided ball position (Fig. 4b).
During backswing. Our study shows that the trends GRF and COP observed at address continues during early backswing, which suggests that the initial setup (GRF Z,L , GRF Z,R , and COP Y ) at address caused by the ball position continued during early backswing, demonstrating a chain effect (Fig. 4a,b). The mediolateral GRF on the right foot (GRF Y,R ) showed a difference between the ball positions during early backswing, where the vertical GRF on the right foot (GRF Z,R ) generally showed peak magnitude. The mediolateral GRF magnitude at the right foot was smaller (less towards the target) for both left-sided and right-sided ball positions than at the reference ball position during early backswing. The trend COM in the M-L direction (COM Y ) observed at address continued throughout the whole backswing, which also demonstrated a chain effect (Fig. 4a,b).   www.nature.com/scientificreports/ der position during the transition for the left-sided ball positions. This seemed to be a strategy employed by the participants to better utilise the X-factor 7,36-38 to absorb greater rotational potential energy during the transition to hit the ball at a longer distance from the club-head at TC when the ball is positioned left. We also found that COM in the M-L direction (COM Y ) observed at address and backswing continued during back-to-downswing transition (Fig. 4a,b), suggesting a chain effect.
During downswing. Many of the kinematic and kinetic variables showed differences during downswing (GRF X,R , A P , A A,L , A A,R , COM Y , and COM X ), whereas the differences between the ball positions observed at address or during early backswing reappeared during downswing (GRF Z,L , GRF Z,R , GRF Y,R , COP Y , A S ). The systematic differences between GRF and COP (GRF Z,L , GRF Z,R , and COP Y ) of the ball positions observed at address and during early backswing reappeared during mid-downswing and continued throughout the impact (Fig. 4a,b). The difference in mediolateral GRF on the right foot (GRF Y,R ) observed during early backswing also reappeared during the later downswing. However, the mediolateral GRF on the right foot during the later downswing showed smaller/greater magnitude (less/more towards the target) in the left-/right-sided ball positions than that at the reference ball position (Fig. 4a,b). The anteroposterior GRF on the right foot (GRF X,R ) at the mid-downswing showed smaller magnitude (less anterior direction) for both the left-sided and right-sided ball positions compared to the reference ball position (Fig. 4a,b).
The difference between the shoulder angles (A S ) of the ball positions observed at back-to-downswing transition continued throughout the downswing. However, the shoulder angle during the early downswing showed a more open position for both the left-sided and right-sided ball positions compared to the reference ball position, and the same trend was observed for the pelvis angle (A P ). The participants rotated their shoulder and pelvis more and created open shoulder and pelvis positions during the early downswing for the right-sided ball position potentially to hit the ball at a shorter distance from the club-head at TC. However, we found that the systematic difference in shoulder angle observed at address reappeared during the mid-downswing, and the same trend was observed in pelvis angle (Fig. 4a,b).
The golf swing during the downswing is executed through linear transition and angular rotation of the body 6,11,12,[39][40][41] . The former can be estimated with the COP in the M-L direction (COP Y ), and the latter with the shoulder and pelvis angles (A S and A P ). COP in the M-L direction during the downswing and impact showed a greater difference in the right-sided ball positions (RF: -23 mm) than that in the left-sided ball positions (LF: + 12.1 mm), whereas the pelvis angle during the downswing and impact showed a smaller difference in the right-sided ball position (RF: -0.2°) than that in the left-sided ball positions (LF: + 1.6°), which demonstrates that the proportion of the linear transition of the body is greater in percentage than the angular rotation of the pelvis in the right-sided ball positions.
Furthermore, the left ankle angle (A A,L ) was more flexed for both the left-sided and right-sided ball positions compared to the reference ball position during the early downswing (Fig. 4a,b). The right ankle angle (A A,R ) was less/more flexed for the left-/right-sided ball positions compared to the reference ball position throughout the whole downswing (Fig. 4a,b).
COM in the M-L direction (COM Y ) shows that the whole-body position is more toward/distant to the target in the left-/right-sided ball positions throughout the whole swing (Fig. 4a,b). This observation also confirms that the chain effect of the ball position continuously leads to systematic changes in the final stage of golf swing. COM in the A-P direction (COM X ) was in a more posterior/anterior position compared to the reference ball position in the left-/right-sided ball positions during late downswing (Fig. 4a,b).
At impact. Most of the kinematic and kinetic variables analysed in our study showed differences during downswing continues at impact (GRF Z,L , GRF Z,R , GRF Y,R , COP Y , A S , A P , A A,R , COM Y , and COM X ), demonstrating the chain effect (Fig. 4a,b). We also found that the anteroposterior COP (COP X ) was in a more posterior/anterior position compared to the reference ball position in the left-/right-sided ball positions at impact (Fig. 4a,b).
Although our study did not directly investigate club-head kinematics or club-ball interaction at impact since the focus was on the swing action influenced by the ball position, another study by Kim et al. (2018) analysed the effect of the M-L ball position on the club-head kinematics and showed that the club path had less/more 'in-out' trajectory at impact in the left-/right-sided ball positions 20 . In our study, the shoulder angle (A S ) is more open during downswing and at impact in the left-sided ball positions. The less 'in-out' trajectory of the club-head at impact in the left-sided ball positions reported in the previous study 20 may be due to a more open shoulder angle during the downswing and impact, whereas the increased 'in-out' in the right-sided ball positions at impact may be due to a more closed shoulder angle during the downswing and impact (Fig. 4c), demonstrating another chain effect.
Statistics and practical applications. Previous studies on golf swing often chose limited discrete swing events for analysis such as address, swing transition, mid-downswing, impact, or the time of the maximum/minimum dependent variable 28,35,[41][42][43] . However, our study employed SPM recently introduced in biomechanics [29][30][31][32][33]44 to analyse the entire golf swing at different discrete swing events. Since golfers use different swing tempos 22,45 , and swing events between golfers do not usually occur at the same time in the time trajectory of swing, it was not ideal to apply SPM to the time trajectory. Therefore, we identified and used 14 sequential swing events for SPM analysis to allow the comparison of dependent variables across swing events rather than time trajectories.
Our findings can potentially be informative to golf coaches and golfers. For example, when a golfer shows a lack of shoulder and pelvis rotations during the downswing, the coach can check whether the ball position is too far to the right and move the ball to the left, and then assess whether the golfer demonstrates improved shoulder