Squat movements have direct relevance to a range of functional movement patterns observed both in sport and daily life (4,16). The barbell back squat is a primary exercise used in many athletic training programs to develop lower body strength and power (16,24). Its use, however, is not just limited to athletic populations but also used extensively in rehabilitation (1). As the squat involves a closed-chain weight-bearing stance, the hamstring muscles are activated to oppose anterior tibial translation forces caused by the quadriceps muscles. This results in less strain being placed on the anterior cruciate ligament (ACL) (19). Therefore, this makes the squat exercise a superior ACL rehabilitation exercise to the leg extension exercise (24).
Depending on the objective of the exercise, squats may be performed as a body weight only movement or “loaded” with additional weight. Regardless, acute and chronic training effects are generally most pronounced on the neuromuscular components and associated connective tissues of the knee, hip, and lower back (7,10). The squat movement can be performed under many variations of technique. Squat depth is one aspect of technique that can vary greatly and is typically prescribed based on the objectives of training. Strength and conditioning coaches typically categorize squats into 3 basic groupings based on depth. These are (a) partial squats represented by knee flexion of 40°, (b) half squats indicated by knee flexion of 70–100°, and (c) deep squats represented by knee flexion of greater than 100°. A bodyweight squat at or below 90° of flexion can be used as a simple measure of total body movement efficiency (16). Poor ankle mobility and a lack of flexibility at the hips are also factors that can affect technique by resulting in the lifter adopting a more pronounced forward lean at the hips. This is often cited as a common technique error (6,8), and it has been speculated that adopting such a position may lead to injury by increasing compressive and shear forces in the lumbar spine (11,14,25). Interestingly, despite the increase in lumbar loading associated with excessive trunk lean, as cited by a number of authors in recent years, the National Strength and Conditioning Association position paper on the squat exercise (3) suggests that lifters restrict the forward movement of their knees, which can, in turn, cause the lifter to lean further forward with the trunk (11,18).
The barbell back squat is a primary exercise in the training programs of athletes who compete in International Weightlifting Federation (IWF) and International Powerlifting Federation (IPF) competitions. Furthermore, the barbell back squat is also one of the 3 competition lifts used in IPF events. Each of these organizations has rules related to competition judging and the use of equipment by participants. One such rule relates to competitors having to wear specified footwear. In the case of the sport of weightlifting, Rule 4.2.1 of the IWF Technical and Competition Rules and Regulations states that “athletes must wear footwear (weightlifting shoes [WS] or boots) to protect their feet and provide stability and a firm stance on the competition platform” ((15), p. 31).
To satisfy the mandate of the IWF, WS are designed with a stiff noncompressible flat sole and a raised heel (approximately 2.5 cm). A metatarsal strap is also used to increase lateral stability by lifting and supporting the transverse arch of the foot. It is suggested that such a design is beneficial to the lifter by limiting flexion and torsion of the sole when under load, without compromising ankle range of motion, thus increasing stability (21). The elevated heel is said to function by allowing lifters to keep their knees forward without being affected by a limited ankle dorsiflexion range of motion, thus making it easier for the lifter to maintain a more upright trunk to minimize external torque and subsequent compressive and shear forces in the lumbar spine (22). Notwithstanding these proposed benefits, the recreational lifter and competitive athlete, who use resistance training to help enhance their performance, typically wear standard sports shoes or other forms of footwear when undertaking resistance training (22). These types of footwear do not have the design characteristics of the WS as described above and would likely provide less stability when under load.
Sato et al. (22) has conducted a preliminary investigation that supports the use of a specialized WS. These researchers used a basic two-dimensional kinematic analysis of three displacement variables to compare the use of WS with standard sports shoes while performing the barbell back squat (60% of 1 repetition maximum [1RM]). Two-dimensional kinematic information is inherently problematic due to errors of perspective, not to mention its inability to comprehensively answer questions regarding multiplanar kinematic and kinetic between-shoe differences (7). Therefore, this study used a more sophisticated three-dimensional (3D) motion capture system to compare the use of WS to standard running shoes (RS) while performing the barbell back squat. Ground reaction force data were also used to track motion of the center of pressure (COP) for each participant. Loads of 50, 70, and 90% of 1RM were prescribed to determine whether any significant differences in key kinematic variables were observed. It was hypothesized that the use of specialized weightlifting footwear would positively influence the selected biomechanical variables being investigated during all barbell back squats. Specifically, it was hypothesized that barbell back squats under all loads would result in less ankle dorsiflexion, less forward trunk lean, and less movement (more stability) of the COP in the WS condition versus the RS condition.
Experimental Approach to the Problem
A within-subject, counter-balanced experimental design was used to determine the impact of the 2 different types of footwear (WS = adipower weightlifting shoe; RS = adidas Response Cushion 22.0) on barbell back squat technique under different %1RM loads (independent variables). Joint kinematics and COP data (dependent variables) were recorded to determine whether there were technique benefits (e.g., optimization of joint kinematics and increased stability). Participants completed 5, 3, and 1 repetition, representing 3 sets, of a barbell back squat pattern using 50, 70, and 90% of their predetermined 1RM, respectively, while wearing each type of footwear (experimental condition). All shoes in this study were new and only used by participants with the same shoe size during this study. A total of 6 sets of the exercise were completed on each of 2 separate testing occasions.
Nine male participants (mean age = 26.4 ± 5.4 years, age range = 18–33 years, height = 1.79 ± 0.08 m, and mass = 84.7 ± 16.1 kg) volunteered to participate in this study. Using a priori power analysis (G*power 3, version 3.1.2) (9) with an effect size of 0.5, power of 0.80, and α level of 0.05, the predicted minimum number of participants would be 8 for detecting within-factor differences using a repeated-measures analysis of variance. All participants were actively engaged in regular weekly resistance training at the time of their participation in this study. Participants also had a minimum of 3 years experience using the barbell back squat in their own training. Participants were informed of the benefits and risks of the investigation before signing an institutionally approved informed consent document to participate in this study. Participants were advised that they could withdraw their consent at any time. All procedures were approved by the Southern Cross University Human Research Ethics Committee (approval number: ECN-13–210).
All testing sessions were conducted at the same time of day (between 1.00 and 4.00 PM) and in the same location. All participants were asked to attend a familiarization session 1 week before testing. At this session, test procedures were demonstrated, and potential experimental risks were identified (e.g., injury). An information sheet detailing the purpose of the study and the nature of participants' involvement was provided. On expressing a willingness to participate in this project, participants were asked to sign informed consent and complete a standard preparticipation health screening questionnaire. Before the first data collection session (test session 2), participants returned at a mutually agreed time (between 1:00 and 4.00 PM) and day to have their baseline measures collected. This required the recording of basic anatomical measures and completion of their 1RM barbell back squat test (test session 1). The 1RM squat test was completed while wearing normal training footwear, as per the procedure outlined by Harman and Garhammer (13).
Squat technique, using an unloaded Olympic bar, was initially assessed by a Certified Strength and Conditioning Specialist before each participant completing their 1RM test; this same coach supervised all subsequent tests during data collection involving the different shoe conditions. Participants presented using a range of depths (from bottom of hamstrings parallel to floor to hamstrings touching the back of the lower leg) at which they normally identified the bottom of their squat movement in their regular training. Notwithstanding the within-subject design nature of this study, squat depth was determined visually and standardized for each participant for the 1RM test and under each of the subsequent shoe conditions. All 1RM tests were conducted in a power rack with depth of squat visually determined as being when the top of the thighs was parallel with the floor and hip crease in line with or slightly below the knee.
Before all test occasions, participants were instructed not to complete any lower body training in the 24 hours immediately before their testing session. Participants were also advised to abstain from consuming a heavy meal or any caffeine-based drinks in the 2 hours before each of the 3 testing sessions and to consume 500 ml of water in the 30 minutes before testing. Participants had to be free from any form of injury and illness to participate in each testing session. Each testing session was separated by a minimum of 48 hours.
On arrival at test session 2, participants were assigned (counterbalanced design) the first pair of shoes (experimental condition) to be worn in that session; the order of shoe was reversed for each participant on the third and last test occasion. Participants were then prepared for testing by the placement of 27 retro-reflective markers on the barbell, head, trunk, pelvis, lower limbs, and feet to define a 10-segment model (modified Plug-In-Gait). Markers identifying the foot segment were placed on the shoes (Figure 1). The 3D coordinates of these markers were captured using 10 high-speed (100 Hz) cameras (Vicon T40-S, Oxford, United Kingdom) and using Nexus software (version 1.8.4, 2013). A Butterworth fourth-order low-pass filter was used to filter marker trajectories (fc = 4 Hz) (12), and 3D marker coordinates were used to calculate the following kinematic variables for each squat in the sagittal plane: peak left and right ankle dorsiflexion angles; minimum left and right knee flexion angles (thigh relative to shank); minimum left and right hip flexion angles (pelvis relative to thigh); minimum left and right thigh inclination angles (thigh relative to the horizontal—determinant of squat depth); and peak forward trunk lean angle (trunk relative to the vertical).
Ground reaction force data were captured (1,000 Hz) during each trial as participants stood with both feet on a single Kistler force plate (Type 9287; Winterthur, Switzerland). The same Nexus software was used for data processing, and ground reaction forces were also filtered (fc = 50 Hz) with a Butterworth fourth-order low-pass filter. The filtered data were used to calculate COP peak displacements in anterior–posterior (A-P) and medio–lateral (M-L) directions.
After participant preparation, each test occasion started with participants completing their normal resistance training session warm-up. They were then required to complete the following sequence of squats: (a) a set of 5 body weight only (i.e., no load) squats followed by 60 seconds rest; (b) a set of 5 squats with an unloaded Olympic bar followed by 2 minutes rest; (c) a set of 5 squats using a load equivalent to 50% of their 1RM followed by a 2–3 minutes rest; (d) a set of 3 squats using a load equivalent to 70% of their 1RM followed by 4 minutes rest; and (e) 1 squat using a load equivalent to 90% of their 1RM. A 5 minute rest period followed this first sequence and was used to allow participants to change into their assigned alternate shoes for the second series of squats. The second series of squats started with the completion of the set of 5 unloaded Olympic bar squats and then continued in the same order as described for the first squat sequence. The bar was located in the high bar position (6) for all repetitions and by all participants.
Verbal feedback was given when participants reached the required squat depth as described above. To standardize squat repetition speed, thereby minimizing unwanted accelerations and variations (23), participants were instructed to squat in time with a digital metronome. In doing so, they executed each full squat movement approximating a tempo equivalent to 1 second down and 1 second up (1-0-1). At the shoe changeover from the first series of squats and after the second and final series, each participant was asked to rate the perceived stability and comfort of each type of shoe. This was performed with the participant sitting on their own and indicating their responses for each shoe by placing a “cross” on a 100 mm visual analog scale. The statement “least comfortable (or stable) shoes I have ever worn” representing the absolute zero point and the statement “most comfortable (or stable) shoes I have ever worn” being the maximum value possible for each question. A similar analog scale was also used for perceived shoe stability.
For each variable of interest, a two-factor (shoe × load) repeated-measures general linear model was then used to determine whether there were main effects of shoe condition (WS and RS) or load condition (50, 70, and 90% 1RM) and whether there were any shoe × load interactions. If any main effects or interactions were detected, then post hoc analysis with Bonferroni adjustment was conducted to identify the significant difference between mean values. Effect sizes using Pearson correlation r were also calculated for all comparisons. Comparisons between shoe comfort and stability were determined using a paired t-test. All statistical analyses were conducted using SPSS version 20 for Windows with an α level set at 0.05.
Initial data analysis involved a day × shoe × load repeated-measures general linear model assessment as data collection procedures were repeated twice. All variables showed no main effect of day and no interactions with day (day × shoe, day × load, or day × shoe × load). Subsequent paired t-tests for all variables were then conducted, and no differences were found between days again for all variables of interest. Therefore, the 2 days were collapsed into 1 session.
Peak Ankle Dorsiflexion Angle
There were no shoe × load interactions found in either the left (p = 0.447, r = 0.22) or right (p = 0.921, r = 0.07) ankles. Significant main effects for shoe were found for peak dorsiflexion of both left (p < 0.001; r = 0.88) and right (p < 0.001; r = 0.85) ankles (Figure 2 and Table 1). Pairwise post hoc comparisons showed that the RS condition exhibited significantly higher dorsiflexion compared to the WS condition in both left and right ankles. There was also a significant main effect of load (%1RM) within the left ankle (p = 0.004; r = 0.53) with post hoc comparisons showing that there was a significant increase in peak dorsiflexion angle from 50 to 90% (p = 0.037) and 70–90% of 1RM (p = 0.034) but no difference between 50 and 70% of 1RM (p = 1.000). Load main effect analysis for the right ankle showed no significance (p = 0.071; r = 0.41).
Minimum Knee Flexion Angle
There were no significant shoe × load interaction effects in either the left (p = 0.527; r = 0.18) or right (p = 0.512; r = 0.18) minimum knee flexion angles (Table 1). Main effects analysis of shoe showed no significance for either the left (p = 0.125; r = 0.36) or right (p = 0.750; r = 0.08), and also no significant effect of load in either the left (p = 0.468; r = 0.19) or right (p = 0.135; r = 0.35) minimum knee flexion angles.
Minimum Thigh Inclination Angle
Shoe x load analysis revealed that there were no significant interactions for either the left (p = 0.465; r = 0.21) or right (p = 0.976; r = 0.03) minimum thigh angles (Table 1). Main effect analysis of shoe indicated no significance for both the left (p = 0.978; r = 0.008) and right (p = 0.862; r = 0.04) minimum thigh inclination angles. Load main effect analysis showed a significant effect for both the left (p = 0.020; r = 0.48) and right thigh (p = 0.045; r = 0.44). Post hoc analysis of the left minimum thigh angle revealed that there were no differences when comparing 50–70% (p = 0.100) and 50–90% (p = 0.094); however, there was a significant increase from 70 to 90% (p = 0.035). For the right minimum thigh angle, post hoc analysis revealed that there were no differences when comparing any of the 3 load conditions (50–70%, p = 0.100; 50–90%, p = 0.133; and 70–90%, p = 0.162).
Minimum Hip Flexion Angle
There were no significant shoe × load interactions for either the left (p = 0.735; r = 0.11) or right (p = 0.90; r = 0.05) minimum hip flexion angles (Table 1). Analysis of shoe main effect indicated no significance for both the left (p = 0.662; r = 0.11) and right (p = 0.788; r = 0.06) minimum hip flexion angles. Load main effect analysis showed a significant effect for the left hip (p = 0.019; r = 0.50) but not for the right hip (p = 0.068; r = 0.44). However, post hoc load analyses for the left minimum hip flexion angle revealed that there were no specific between-load differences (50–70%, p = 0.334; 50–90%, p = 0.059; and 70–90%, p = 0.103).
Peak Trunk Lean Angle
There was no significant shoe × load interaction for the peak trunk lean angle (p = 0.308; r = 0.26). Shoe main effect analysis indicated no significance for the peak trunk lean angle (p = 0.502; r = 0.16). However, a significant load main effect was found (p = 0.006; r = 0.58) (Table 1). Post hoc analyses for the peak trunk lean angle found no difference between 50 and 70% (p = 0.213), but there was a significant increase from 50 to 90% (p = 0.022) and from 70 to 90% of 1RM (p = 0.016).
Peak Center of Pressure Deviations
Analysis of peak COP (A-P) deviation showed no significant interaction (p = 0.344; r = 0.25); however, there was a significant main effect of shoe (p = 0.010; r = 0.57; Figure 3) and load (p = 0.038; r = 0.42). Post hoc analyses revealed a significantly larger peak COP (A-P) movement in the WS condition compared with the RS condition (p = 0.010). Analysis of the effect of load showed that there were no differences between 50 and 70% (p = 1.000) or 50 and 90% of 1RM (p = 0.273); however, there was a significant reduction in peak COP (A-P) from 70 to 90% of 1RM (p = 0.049). Analysis of peak COP (M-L) deviation showed no significant interaction (p = 0.452; r = 0.20) and no main effects of shoe (p = 0.623; r = 0.12; Figure 3) or load (p = 0.437; r = 0.22) (Table 1).
Perceived Shoe Comfort and Stability
Paired t-tests indicated that participants perceived the RS to be significantly more comfortable (p < 0.032; mean rating = 80.4 ± 11.9) than the WS (mean rating = 70.2 ± 14.9), whereas the WS were perceived to be significantly more stable (p < 0.000; mean rating = 89.5 ± 7.6) than the RS (mean rating = 65.9 ± 15.5).
This project aimed to establish whether specific WS significantly affect technique during a barbell back squat when compared with a standard sport RS under different %1RM loads. Although limited research has been conducted previously (22), this study used more sophisticated methodology (3D motion capture) to determine differences in key kinematic measures. It also used ground reaction force data to determine displacement measures as indicators of stability in A-P and M-L directions.
Confirming our hypothesis, this research revealed a significant main effect for shoe in peak left and right dorsiflexion angles with the heel-elevated WS exhibiting significantly lower dorsiflexion angles compared with the RS. Because of the solid construction of the elevated heel and the absence of a soft midsole in the WS, this finding is not surprising. However, unlike previously reported findings (22), our analyses revealed no other effect of shoe condition on any of the selected key kinematic variables. Consequently, our hypothesis that any reduction in peak dorsiflexion in the WS would also lead to a reduction in peak forward trunk lean was not confirmed.
It must be noted that Sato et al. (22) did not directly assess trunk lean, and that their conclusions regarding the effect of shoes on dorsiflexion and trunk lean are limited by this. Nonetheless, it is reasonable to postulate that by simple geometry, their measures of A-P bar and hip displacement could be deemed to reflect trunk lean, which is what informed our original hypothesis. Another possible reason for not observing an effect of shoe on trunk lean in this study may be that participants had sufficient ankle dorsiflexion range of motion to cope with the RS condition (as reflected by the increase in peak dorsiflexion). This may have enabled them to maintain a forward-knees position, a technique advised by Fry et al. (11) and List et al. (18), to keep the load displacing vertically over their base of support (4) without needing to increase hip flexion or forward trunk lean. Although knee position relative to the feet was not calculated in this study, it is reasonable to postulate that the greater peak dorsiflexion angle was likely to compensate for a lower and softer heel wedge in the RS condition to result in a similar knee position. Furthermore, the similarities in the kinematic measures further up the chain indicate some between-shoe consistency in squat technique, at least in sagittal kinematic minima and maxima. Future studies, however, should investigate the impact of dorsiflexion capacity on technique during squats in different footwear.
Our analyses have also shown that peak COP displacements, typical indicators of stability and balance (17,20), differed between shoe conditions for these participants with a significantly larger COP (A-P) displacement in the WS condition compared to the RS condition. Even though there was no between-shoe difference in peak M-L COP deviation, these findings mean that the hypothesis that the WS condition would lead to less overall COP movement (greater stability) was rejected. Nonetheless, it is worth noting that this finding was not supported by the perception of the study participants. The results of the visual analog scale assessments on both days showed that the cohort rated the WS condition as significantly more stable than the RS condition. One reason for this may be the assumption that reduced amplitude of displacement in the COP is not necessarily an indicator of greater balance, stability, or control. In fact, researchers are now starting to postulate that these gross measures of COP excursion fail to provide temporal information that can help to explain strategy and control (2). It may be that participants, who perceived the WS condition as more stable, were more confident and therefore allowed themselves to translate more in the A-P direction to manipulate the location of the load in relation to the base of support. For instance, Croft et al. (5) demonstrated that COP displacement was more constrained in standing balance tasks where participants were on compliant (unstable) vs. solid surfaces. Any future investigations of balance or stability effects of footwear in squatting movements should therefore incorporate nonlinear approaches that quantify temporal motor control strategies and responses (2).
Although there were no hypotheses developed for this analysis, it is interesting that a load effect was observed for COP deviations in this study. Although there was no significant difference in peak COP (A-P) displacement between 50 and 70 and between 50 and 90% 1RM, there was a significant reduction in peak COP (A-P) displacement from 70 to 90% of 1RM. It is reasonable to postulate that a feed-forward mechanism was used by participants to constrain movement in the most difficult load condition and perhaps there is a load threshold, yet to be elucidated, that alters this.
Overall, there were no significant shoe effects in all key selected kinematic measures, except for peak ankle dorsiflexion angles. Although peak trunk lean was consistent between shoe conditions, conclusions regarding potential lumbar loading remain tenuous as no joint kinetics were measured or calculated to provide more direct evidence. Future studies should, therefore, incorporate a kinetic analysis. Nonetheless, links between kinematics and kinetics are well established during squatting and lifting tasks (10,11,14,24), and so the findings of this study certainly provide a sound basis for more work. It is also worth considering that greater peak displacement in COP (A-P) deviation in the WS condition, albeit while constraining peak angular displacements, may be an indication that the temporal between-shoe movement patterns were different. As a result, future work with the current data set may be warranted, using more comprehensive techniques such as principal component and support vector machine analyses. Such an approach may highlight more subtle shoe effects that are easily missed by traditional discrete variable analyses such as this study. It must also be acknowledged that conclusions in this study are limited by a cohort of 9 participants. However, statistical findings and inferences were strengthened by the study design that incorporated a repeat of measures across 2 days and across 3 load conditions. Having no effect of day, for instance, enabled the data to be collapsed into a shoe × load design, effectively doubling the size of the data set.
This research has been driven by practices observed in the training environment regarding the use of different types of footwear. To date, there is limited evidence for support of this practice and any potential beneficial effects on lifting performance or reducing injury risk. As is evident by the discussion of findings in this study, the scope for studying the effects of footwear on squat movements should be expanded substantially. Athletes, coaches, patients, and therapists undertake squats for many reasons and by using varying techniques and footwear. However, it is our contention that the results of this study question the notion that overall technique, and therefore injury risk, may be substantially affected by footwear typically used in weightlifting environments. Notwithstanding this, it is important that further studies, such as those suggested in this article, are undertaken to provide a clearer understanding of the benefits, or otherwise, of this type of footwear to users.
The authors acknowledge the support of Adidas (a.i.t) who provided the shoes used in this study.
1. Abelbeck KG. Biomechanical model and evaluation of a linear motion squat type exercise. J Strength Cond Res 16: 516–524, 2002.
2. Baltich J, von Tscharner V, Zandiyeh P, Nigg BM. Quantification and reliability of center of pressure movement during balance tasks of varying difficulty. Gait Posture 40: 327–332, 2014.
3. Chandler T, Stone M. The squat exercise in athletic conditioning: A position statement and review of the literature. NSCA J 13: 51–58, 1991.
4. Clark DR, Lambert MI, Hunter AM. Muscle activation in the loaded free barbell squat: A brief review. J Strength Cond Res 26: 1169–1178, 2012.
5. Croft JL, von Tscharner V, Zernicke RF. Movement variability and muscle activity relative to center of pressure during unipedal stance on solid and compliant surfaces. Motor Control 12: 283–295, 2008.
6. Donnelly DV, Berg WP, Fiske DM. The effect of the direction of gaze on the kinematics of the squat exercise. J Strength Cond Res 20: 145–150, 2006.
7. Escamilla RF, Fleisig GS, Lowry TM, Barrentine SW, Andrews JR. A three-dimensional biomechanical analysis of the squat during varying stance widths. Med Sci Sports Exerc 33: 984–998, 2001.
8. Fairchild D, Hill B, Ritchie M, Sochor D. Roundtable: Common technique errors in the back squat. Natl Strength Cond Assoc J 15: 20–27, 1993.
9. Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39: 175–191, 2007.
10. Flanagan SP, Salem GJ. Bilateral differences in the net joint torques during the squat exercise. J Strength Cond Res 21: 1220–1226, 2007.
11. Fry AC, Smith JC, Schilling BK. Effect of knee position on hip and knee torques during the barbell squat. J Strength Cond Res 17: 629–633, 2003.
12. Hadi G, Akkus H, Harbili E. Three-dimensional kinematic analysis of the snatch technique for lifting different barbell weights. J Strength Cond Res 26: 1568–1576, 2012.
13. Harman E, Garhammer J. Administration, scoring, and interpretation of selected tests. In: Baechle T.R., Earle R.W., eds. Essentials of Strength Training and Conditioning (3rd ed.). Champaign, IL: Human Kinetics, 2008. pp. 249–292.
14. Hart DL, Stobbe TJ, Jaraiedi M. Effect of lumbar posture on lifting. Spine 12: 138–145, 1987.
15. International Weightlifting Federation. IWF Technical and Competition Rules & Regulations 2013–2016. Budapest, Hungary: The International Weightlifting Federation, 2013.
16. Ktitz M, Cronin J, Hume P. The bodyweight squat: A movement screen for the squat patter. Strength Cond J 31: 76–85, 2009.
17. Lafond D, Duarte M, Prince F. Comparison of three methods to estimate the center of mass during balance assessment. J Biomech 37: 1421–1426, 2004.
18. List R, Gülay T, Stoop M, Lorenzetti S. Kinematics of the trunk and the lower extremities during restricted and unrestricted squats. J Strength Cond Res 27: 1529–1538, 2013.
19. Neumann DA. Kinesiology of the Musculoskeletal System: Foundations for Rehabilitation (2nd ed.). St. Louis, MO: Mosby/Elsevier, 2010. pp. 560–562.
20. Prieto TE, Myklebust JB, Hoffmann RG, Lovett EG, Myklebust BM. Measures of postural steadiness: Differences between healthy young and elderly adults. IEEE Trans Biomed Eng 43: 956–966, 1996.
21. Rippetoe M, Kilgore L. Starting Strength: Basic Barbell Training (2nd ed.). Witchita Falls, TX: The Aasgaard Company, 2007. pp. 60–61.
22. Sato K, Fortenbaugh D, Hydock DS. Kinematic changes using weightlifting shoes on barbell back squat. J Strength Cond Res 26: 28–33, 2012.
23. Sato K, Heise GD. Influence of weight distribution asymmetry on the biomechanics of a barbell back squat. J Strength Cond Res 26: 342–349, 2012.
24. Schoenfeld BJ. Squatting kinematics and kinetics and their application to exercise performance. J Strength Cond Res 24: 3497–3506, 2010.
25. Swinton PA, Lloyd R, Keogh JL, Agouris I, Stewart AD. A biomechanical comparison of the traditional squat, powerlifting squat, and box squat. J Strength Cond Res 26: 1805–1816, 2012.