Fitness professionals examine and treat persons with pain in the posterior neck, mid-thoracic spine, and shoulder because of “rounded shoulders” posture. During relaxed standing or unsupported sitting, the scapulae of persons with “rounded shoulders” are typically depressed, downwardly rotated, and protracted with internal rotation and anterior tilt (18). Observable changes in the static position of the scapula and scapular movement patterns relative to the thoracic cage have been termed “scapular dyskinesis” (16).
During weight training activities, scapular dyskinesis can result from an imbalance between upper body pressing and pulling performance (4). Emphasis on pressing movements in upper extremity resistance training increases the likelihood the scapulae are protracted, internally rotated, and anteriorly tilted from a strength imbalance between the scapular protractors and retractors (8). Furthermore, an imbalance between the activation of upper fibers (UT) and lower fibers (LT) of the trapezius has been shown to contribute to symptomatic subacromial impingement, a condition rehabilitation professionals often treat (20).
When prescribing exercise intended to reduce scapular dyskinesia, fitness professionals focus on pulling exercises designed to strengthen the scapular retractors. One option is the inverted body weight row, an exercise intended to enhance pulling performance. It recruits a variety of scapular retractors and vertebral stabilizers, which are antagonists to the shoulder girdle protractors. An inverted row involves a supine posture while reaching to grip a bar extending horizontally above the person. While keeping the feet on the floor and the torso rigid, the body is pulled vertically until maximum elbow flexion is achieved. The elbows are then extended until the person returns to the starting position.
Fenwick et al. (12) described EMG activation of torso and hip muscles when healthy subjects performed an inverted row. The inverted row was found to generate a modest lumbar spine load compared with the standing bent-over and one-armed cable rows. The authors suggested the inverted row would be the preferred exercise for clients with low back pain because of low spinal loads. Nevertheless, with the exception of the latissimus dorsi, Fenwick et al. did not investigate EMG activation of scapular and shoulder muscles, which are often targeted by rehabilitation professionals who desire to strengthen scapular muscles in clients with “rounded shoulders.” Therefore, the purpose of this study was to quantify muscle activation of 9 muscles using surface EMG analysis during 4 inverted row body weight exercises using a portable pull-up device. The 4 exercises were: (a) both feet on the ground with forearms pronated; (b) both feet on the ground with forearms supinated; (c) single-limb support with forearms pronated; and (d) single-limb support with forearms supinated. We hypothesized muscle EMG activation would be significantly greater in the single-limb support conditions when compared with double-limb support because of the assumed demand for additional core muscle activation to help maintain postural stability on a reduced support surface. Furthermore, we hypothesized muscle activation would be significantly greater in the BB with forearms in the supinated position as compared to the pronated position. This is consistent with what has been previously found during a chin-up (22).
Experimental Approach to the Problem
The inverted row targets spinal stabilizers and shoulder complex muscles that oppose upper body pressing actions (8). This exercise can be performed with a pronated or supinated grip and double- or single-limb weight-bearing. We examined 9 muscles recruited during 4 different inverted row exercise conditions including the following: (a) posterior deltoid (PD), (b) latissimus dorsi (LD), (c) biceps brachii (BB), (d) lower trapezius (LT), (e) upper trapezius (UT), (f) lumbar multifidus (LM), (g) middle trapezius (MT), (h) lumbar thoracis (LTh), and (i) rectus abdominis (RA). EMG data were analyzed with Delsys software by processing the root-mean-square (RMS) algorithm and normalizing values to a maximum voluntary isometric contraction (MVIC).
Independent variables examined were the 4 exercise conditions. The dependent variables were normalized EMG values (% MVIC) of the 9 muscles. The questions we sought to answer were as follows: (1) “Does a supinated grip result in greater activation of the BB than a pronated grip?” and (2) “Does single-limb support result in greater activation in all muscles studied when compared with double-limb support?”
The study consisted of 26 subjects ranging in age from 22 to 39 years. Subject demographics are summarized in Table 1. Participants were recruited using a convenience sample from Mayo Clinic College of Medicine, School of Health Sciences. To be eligible to participate in the study, volunteers needed to be between 20 and 39 years of age.
To be included in the study, each subject demonstrated normal active range of motion of the shoulder, hip, knee, ankle, and foot as assessed visually by one of the investigators. Volunteers reporting a history of the following conditions were excluded from the study: (a) previous subluxation, dislocation, or fracture; (b) a history of joint instability, tendinitis, bursitis, impingement, adhesive capsulitis, neurovascular complications, or any condition that limited physical activity for greater than 2 days over the last 6 months; and (c) current complaints of neuromuscular pain, numbness, or tingling in the upper extremity, lower extremity, and back. A sample size of 22 subjects was required to detect a mean difference in EMG recruitment of 10% MVIC (effect size = 0.20) between conditions with a statistical power (1-β) equal to 0.80 at α = 0.05 (11). The procedures of this study were approved by the Mayo Institutional Review Board, Mayo Clinic, Rochester, MN, USA. Each participant was informed of the benefits and risk of the investigation before signing an institutionally approved informed consent document to participate in the study.
Raw EMG signals were collected using Bagnoli DE 3.1 double-differential surface EMG sensors. The sampling frequency used was 1,000 Hz. Sensor contacts were made from 99.9% pure silver bars 10 mm in length, spaced 10 mm apart, and encased with preamplifier assemblies measuring 41 × 20 × 5 mm. Preamplifiers had a gain of 10 V/V. The combined preamplifier and main amplifier permitted a gain from 100 to 10,000. The common mode rejection ratio was 92 dB at 60 Hz, and input impedance was >1,015 Ω at 100 Hz. Raw EMG signals were processed with EMG works Data Acquisition and Analysis software (Delsys Inc., Boston, MA, USA).
Data were collected in a research laboratory at Mayo Clinic in Rochester, MN. Before data collection, subjects selected note cards with numbers corresponding to each exercise. The investigator documented the order of the exercises for each subject. This method of exercise selection was used to randomize the order of exercises performed. Subjects were instructed to wear clothing permitting correct placement of the EMG electrodes. The skin over the muscle belly was prepared by shaving hair from the vicinity and cleansing with isopropyl alcohol wipes. Surface electrodes were attached with adhesive interfaces (Delsys Inc.) and secured with 3M Transpore medical tape (Micropore Plus, St. Paul, MN, USA). A common ground electrode was placed over the medial malleolus of the right lower extremity. Electrodes (Figure 1) were placed superficially and parallel to the direction of the muscle fibers on the subject's right side using established techniques for each of the 9 muscles (5). Wires from the electrodes were connected to an input module attached to a gait belt fastened around the subject's waist.
Manual Muscle Test Procedures
Subjects then performed a series of MVICs to set the gain on the Delsys EMG instrumentation. Manual muscle test procedures were described in a muscle-testing textbook (13). For each muscle test, examiners provided verbal encouragement to help the subject produce and maintain a consistent effort. After each MVIC procedure, subjects were asked if the test was performed at maximum effort. If not, the MVIC was repeated.
For the RA manual muscle test, the participant was supine with hands clasped behind the occiput. With hips and knees in extension, the subject performed thoracoabdominal flexion so the scapulae were elevated from the table surface (13). The examiner asked the subject to maintain the position, whereas manual resistance was placed over the anterior sternum in the direction of trunk extension. Another investigator stabilized the subject's lower extremities by grasping the subject's ankles.
To test the LTh and LM, the subject performed prone trunk extension with hands clasped behind the occiput. An investigator stabilized the lower extremities by grasping the malleoli. With trunk extension, the sternum was no longer in contact with the support surface. At end range of trunk extension, the examiner applied manual resistance to the upper thoracic region in the direction of trunk flexion (13).
For the UT muscle test, the subject was seated with hands relaxed in his or her lap. The examiner instructed the subject to shrug the shoulders and hold them in an elevated position. The examiner stood behind the subject and applied maximum resistance to the lateral acromion in an attempt to depress the shoulder (13). Another investigator stood on the subject's left side and stabilized the shoulder.
To test the MT, the subject was positioned prone and instructed to abduct the right shoulder to 90° with the right elbow flexed to 90°. The examiner stood at the subject's side and stabilized the contralateral scapular region. The subject adducted/retracted the scapula and horizontally abducted the arm to 90°. The examiner applied maximum resistance to the subject's distal arm in the direction of scapular protraction and shoulder horizontal adduction (13).
For the LT, the subject was positioned prone with the right upper extremity in 145° of abduction and in line with the lower fibers of the trapezius. The forearm was placed in neutral position with the right thumb pointed toward the ceiling. The examiner stood on the right side with the hand placed on the subject's distal arm. The subject raised the arm from the table to ear level and held this position, whereas the examiner applied resistance straight downward toward the floor in the direction of shoulder extension (13).
To test the LD, the subject was positioned prone with the head turned to the right and the shoulder elevated to the level of the chin. The examiner stood on the right side of the table and grasped the subject's forearm above the wrist with both hands. The subject was instructed to depress the shoulder caudally approximating the shoulder and pelvis. Upon completion of the shoulder depression, the examiner attempted to push the arm upward toward the head (15).
For the PD, the subject was positioned prone with the right upper extremity abducted to 90° with the arm off the edge of the support surface and the elbow extended with the forearm pronated. The examiner stood on the right side with the right hand over the distal aspect of the arm just proximal to the elbow joint. The subject was instructed to extend the arm raising it off the table while keeping the elbow in extension. The examiner applied resistance to the distal arm in the direction of shoulder flexion (13). The examiner's left hand was placed over the subject's left shoulder to assist with trunk stabilization.
To test the BB, the subject was seated with the right forearm supinated and elbow flexed to about 125°. The examiner's right hand was placed at the distal aspect of the subject's forearm and the left hand placed over the anterior shoulder. With the right hand, the examiner attempted to extend the subject's elbow, whereas the left hand applied a counterforce to resist upper arm movement (13).
Inverted Row Exercises
Before performing the 4 exercises, participants were given verbal instruction and a demonstration of each exercise. Subjects practiced each of the exercises under the supervision of an examiner to ensure proper form. The participants then performed 3 repetitions of each exercise in random order over a 10-second time period. To reduce the potential for fatigue, participants were provided a 2-minute rest break between each exercise. A metronome was set to 50 b·min−1 to standardize the rate of movement among subjects. This rate was established before data collection via a pilot study because it allowed participants to complete 3 repetitions of each exercise within a 10-second time period.
For exercise one, the inverted row was performed with both feet on the floor and a pronated grip (Figure 2). The subject began the exercise with the lower extremities extended and the spine in a neutral position. Subject's arms were flexed to grasp the portable pull-up device known as the WorkHorse (Work Horse Fitness Inc., Colony, TX, USA). Upon an investigator's command, the subject lifted the upper torso away from the floor by flexing the elbows and elevating the upper torso toward the horizontal bar. The subject then lowered the torso until the elbows were straight.
Exercise 2 was performed similarly to exercise 1. The subject performed the exercise with both heels on the floor with hip and knee extension (Figure 3). Unlike exercise #1, the subject grasped the horizontal bar with the palms facing the face (supinated). Upon command, the subject lifted the upper torso away from the floor by flexing the elbows and elevating the upper torso toward the horizontal bar. The subject then lowered the torso until the elbows were straight.
Exercise 3 was performed identical to exercise 1 except the left lower extremity was non–weight-bearing (NWB) with the hip and knee joints in extension. The right lower extremity was weight-bearing with the heel in contact with the floor (Figure 4).
For exercise 4 (Figure 5), the subject performed the inverted row exercise identical to exercise 2 except the left lower extremity was NWB with the hip and knee joints in extension. The right lower extremity remained in contact with the floor.
The dependent variable was normalized peak EMG activity (% MVIC) for each of the 9 muscles. MVICs were collected to normalize data and permit meaningful comparisons among study subjects. Raw EMG data collected during the tests were band-pass filtered between 20 and 450 Hz and subsequently processed with a root-mean-square algorithm using moving windows with 125-milliseconds time constants. EMG data collected during the inverted rows were normalized to each muscle's respective MVIC trials and therefore expressed as a percentage of the MVIC (% MVIC). Peak activation for each muscle was calculated from the normalized data using a 200-milliseconds window about the peak amplitude displayed during the 10-second exercise interval. To assist with classification of muscle recruitment from very high to low, we used a scheme initially described in 1992 (7). The categorization is as follows: very high is >60% MVIC, high is 41–60% MVIC, moderate is 21–40% MVIC, and low is 0–20% MVIC.
Descriptive statistics of EMG activation during 4 versions of the inverted row exercise were calculated and displayed (Table 2). Data distributions were examined for normality using Shapiro–Wilk tests. Nearly 90% of the distributions did not violate the assumption that data were normally distributed (p > 0.05) and therefore we used parametric tests to examine differences in muscle recruitment across the exercises. Data from each muscle were examined separately with a repeated measures analysis of variance (ANOVA) at α = 0.05. If the assumption of sphericity was violated in the repeated measures analysis, then Greenhouse-Geisser corrections for degrees of freedom were applied. Post hoc comparisons of the magnitudes of EMG activation across exercises for statistically significant ANOVAs were conducted with Bonferroni corrections for multiple comparisons. Data were analyzed with IBM SPSS 21.0 software (IBM Corp., Armonk, NY, USA).
There were no statistically significant differences in muscle activation between single- and double-leg WB in all muscles analyzed. Four muscles (BB, LD, LT, and PD) demonstrated very high (>61% MVIC) EMG activation during all 4 exercise conditions. Three muscles (UT, MT, and LM) demonstrated high (41–60% MVIC) activation, whereas 2 muscles (LTh and RA) demonstrated moderate (21–40% MVIC) activation. We found statistically significant differences in EMG recruitment (% MVIC) in the LD between the supinated double-leg WB (94.0 ± 21.8% MVIC) and pronated double-leg WB (79.0 ± 25.4% MVIC) with a p-value = 0.001 (Figure 6). We also found statistical significance in the UT between the pronated single-leg WB (67.0 ± 28.8% MVIC) and supinated single-leg WB (56.7 ± 29) with a p-value = 0.007 (Figure 7). No statistically significant differences were observed in muscle activation between single- and double-leg WB in any muscles analyzed for both the pronated grip and supinated grip.
We hypothesized muscle EMG activation would be significantly greater in the single-limb support conditions as compared to double-limb support because of the assumed demand for additional core muscle activation to help maintain postural stability on a reduced support surface. There were no statistically significant differences in muscle activation between single- and double-leg WB in all muscles analyzed during the inverted row exercise. Furthermore, we hypothesized muscle activation would be significantly greater in the BB with forearms in the supinated position as compared to the pronated position. Our data did not demonstrate a difference in BB muscle recruitment in the supinated grip when compared with the pronated grip position.
To our knowledge, previous research has not reported normalized EMG activity of the posterior shoulder and scapular muscles during inverted row exercises. The aim of this study was to quantify muscle recruitment of 9 muscles with surface EMG analysis during 4 inverted row body weight exercises using a portable pull-up device. Researchers have proposed the threshold value for muscle strength gains during exercise requires EMG activation greater than 50–60% MVIC (1,3). Therefore, EMG signal amplitude (% MVIC) was used as a scheme to discuss the relative exercise intensity of the back, abdomen, and scapulohumeral muscles of the right side during 4 inverted row exercises. EMG analysis does not provide a direct measure of muscle strength; instead, it merely quantifies the neuromuscular activity underneath the surface electrode. However, normalized muscle recruitment can be used to determine which exercise patterns produce highest external demand upon a muscle.
The posterior deltoid (14) functions to extend and externally rotate the arm at the glenohumeral joint. Neumann (18) described the importance of the PD during coupling of elbow flexion and shoulder extension. This movement pattern allows the BB to remain at a biomechanical advantage for producing elbow-flexion torque. During the current study, activation of the posterior deltoid was recorded at a very high level. Schoenfield et al. (19) had subjects perform seated rows to fatigue on a reverse fly machine with the resistance set to 75% of body mass and also reported a very high level of PD activation. Andersen et al. (2) also found very high-peak EMG activation of the PD during 3 repetitions of an upright row using 100% of a subject's 8 repetitions maximum. Each reported very high recruitment at levels appropriate for strengthening the PD (1,3).
Actions of the LD (14) include extending, internally rotating, and adducting the arm at the glenohumeral joint. During the current study, activation of the LD was very high and comparable to the value generated during a chin-up with body weight resistance (22). Likewise, Fenwick et al. (12) found the inverted row using chains with handles suspended from the ceiling and body weight resistance produced very high activation in the LD. All 3 studies reported EMG activation levels at very high range and at levels conducive to strengthening. It appears for individuals unable to perform the chin-up, the inverted row is a good beginning exercise for strengthening the LD.
The UT and LT both function as upward rotators of the scapula. The UT also elevates the scapula, whereas the LT produces scapular depression (14). Both are essential for stabilization of the scapulothoracic joint during upper extremity movement (17). In the present study, activation for the UT was high and very high for the LT. Youdas et al. (23) also reported a very high activation level for the UT when performing scaption with a dumbbell in standing. They also recorded high LT activation with the same exercise. Similarly, LT activation during a chin-up using body weight resistance (22) was found to be high.
Smith et al. (20) examined the ratio of UT to LT activation in persons with symptomatic subacromial impingement (SAI). During scapular plane humeral elevation in patients with SAI, they found the mean UT: LT ratio was 3.15, whereas asymptomatic subjects produced a mean UT: LT ratio of 1.80. This indicates strengthening the LT can be beneficial in normalizing scapular movement and symptoms of SAI. During the inverted row, the LT displayed muscle recruitment conducive to strengthening. Furthermore, during the inverted row, the shoulder is able to maintain a position that tends to be nonpainful for people with SAI (0–90° shoulder flexion and neutral internal/external rotation). When compared to shoulder position during a chin-up or scaption, it becomes clear that the inverted row has great potential as a therapeutic exercise in persons with SAI.
The BB (14) produces flexion and supination of the forearm at the elbow joint and flexion of the arm at the glenohumeral joint. We hypothesized that the BB would have higher activation during an inverted row with a supinated vs. pronated grip. However, during the current study, activation of the BB was similar in all 4 conditions with very high activation ranging. Youdas et al. (22) reported BB activation was very high during the chin-up (supinated grip) and the pull-up (pronated grip).
The MT acts to (14) retract the scapula and stabilize the scapulothoracic joint. During the current study, the activation of the MT was moderate. This was lower than we had expected based on findings by other investigators whereby MT activation during scaption, horizontal adduction, or unilateral rowing demonstrated recruitment levels appropriate for strengthening. Youdas et al. (23) found MT activation during scaption to be very high. Ekstrom et al. (9) also found much higher activation during high-intensity exercises. Subjects performed a unilateral row and horizontal adduction with external rotation in the prone position using 85–90% of their 5 repetitions maximum. This elicited very high activation for horizontal adduction and the unilateral row. Likely, the discrepancy between values reported in this study and those of other investigators is related to the amount of resistance and line of force. The position of the body during an inverted row naturally reduces the amount of weight a person must lift as compared to the full body weight of a chin-up. Both the unilateral row and horizontal adduction allow for greater glenohumeral extension, thus encouraging greater scapular retraction and MT activation.
The RA acts as a primary trunk flexor, while bilateral contraction of either the LTh or LM produces trunk extension (14). Unilateral contraction of the LTh produces ipsilateral lateral flexion of the vertebral column and unilateral contraction of the LM results in contralateral rotation and lateral flexion of the vertebral column (14). Activation of the RA during this study was moderate. Other investigators have found greater magnitudes during body weight exercise. For example, Ekstrom et al. (10) reported high activation during a body weight prone bridge, whereas Czaprowski et al. (6) reported a high value of the RA during a prone bridge on a Swiss ball. During the current study, the activation of the LTh and LM were moderate. Likewise, Ekstrom et al. (10) reported moderate activation for the LTh and high activation for the LM during various bridging and quadruped exercises. Moreover, the core muscles—RA, LTh, and LM—investigated in this study produced moderate-to-high muscle activation. According to McGill (17), this level of activation fits the function of core musculature during dynamic trunk movements because their role in spinal stabilization requires frequent and extended periods of recruitment. Therefore, activation of core musculature at levels observed during inverted row exercises shows promise as a therapeutic exercise. It should also be noted that performing an inverted row allows the lumbar spine to maintain a more neutral position and has been found to exert fewer compressive forces on the lumbar spine than other rowing exercises (17).
We hypothesized that there would be increased EMG activation in the single-limb support vs. the double-limb support. However, data did not support this hypothesis. The core muscles—RA, LTh, and LM—did not exhibit higher levels of activation in the single-limb support condition. Perhaps, the use of single-limb support posture did not provide appropriate challenges to the neuromuscular system in well-trained participants (21). Anecdotally, subjects commented that they felt more activation in the gluteal and hamstring muscles during single-limb support than double-limb support, suggesting these muscles, and not the RA, LTh, and LM, likely played a greater role in stabilizing the body when the base of support was decreased. Future research quantifying lower extremity activation during an inverted row would shed light on this topic.
There are a variety of limitations in this study. As with all surface EMG electrode systems, there is the potential for skin artifact and cross-talk. However, the effect of these was reduced by cleansing the skin with alcohol, shaving excess hair, and reinforcing electrodes with tape (3M Transpore) as needed. We also standardized electrode placement based on techniques described by Criswell (5). Additionally, the external validity of our investigation could be limited because the study included a small sample size, and the results apply to young, healthy, and active individuals and not necessarily to the general population or others with shoulder pathology. Lastly, the device for performing the inverted row exercise was nonadjustable. This resulted in varying starting positions for subjects. Further research regarding inverted row exercises should include a therapeutic population to examine its effects on function. Likewise, a broader age range and larger sample size would increase external validity. It would also be beneficial to investigate how or if muscle activation differs when adding external weight (e.g., weighted vest or backpack) to the participants or performing the inverted row on unstable surfaces such as a BOSU ball.
Athletic trainers, personal trainers, strength and conditioning coaches, and physical therapists prescribe various strengthening exercises to their clients for rehabilitation and strengthening of the scapulohumeral, back, and abdominal muscles. Therefore, knowing which inverted row condition provides the greatest strengthening effect is beneficial when targeting specific muscles. Based on results of the current study, each of the 4 inverted row exercises displayed very high to high muscle activation at a level appropriate for strength training in the BB, LD, LT, UT, and PD.
We used a commercially available portable pull-up device that resembled a sawhorse and body weight resistance to assess torso and posterior shoulder girdle muscle activation during the inverted row. However, fitness professionals have several equipment options to choose from when instructing clients how to strengthen spinal stabilizers and posterior chain muscles using the inverted row. In fitness facilities, clients or athletes can use a barbell horizontally mounted on a standard barbell rack or Smith machine or TRX suspension straps. At home instead of a barbell, individuals may substitute a sturdy wooden dowel or pipe positioned horizontally between 2 kitchen chairs. Alternatively, a desk or table with a solid surface to grip can be substituted for a bar. The exerciser lies supine beneath the dowel, pipe, or desk and reaches vertically to grasp the bar or desktop. Body weight can be increased by using a weighted vest or barbell plates secured within a backpack.
Funding for this project came from the Mayo Program in Physical Therapy, College of Medicine, Mayo Clinic, Rochester, MN, USA. The authors disclose that they have no professional relationship with any of the materials or equipment used during this study. The results of the present study do not constitute endorsement of the devices by the authors or the National Strength and Conditioning Association.
1. Andersen LL, Magnusson SP, Nielsen M, Haleen J, Poulsen K, Aagaard P. Neuromuscular activation in conventional therapeutic exercises and heavy resistance exercises: Implications for rehabilitation. Phys Ther 86: 683–697, 2006.
2. Andersen LL, Kjaer M, Andersen CH, Hansen PB, Zebis MK, Hansen K, Sjøgaard G. Muscle activation during selected strength exercises in women with chronic neck muscle pain. Phys Ther 88: 703–711, 2008.
3. Ayotte NW, Stetts DM, Keenan G, Greenway EH. Electromyographical analysis of selected lower extremity muscles during 5 unilateral weight-bearing exercises. J Orthop Sports Phys Ther 37: 48–55, 2007.
4. Baker DG, Newton RU. An analysis of the ratio and relationship between upper body pressing and pulling strength. J Strength Cond Res 18: 594–598, 2004.
5. Criswell E. Cram's Introduction to Surface Electromyography (2nd ed.). Sudbury, MA: Jones and Bartlett Publishers, 2011.
6. Czaprowski D, Afeltowicz A, Gębicka A, Pawłowska P, Kędra A, Barrios C, Hadała M. Abdominal muscle EMG-activity during bridge exercises on stable and unstable surfaces. Phys Ther Sport 15: 162–168, 2014.
7. DiGiovine NM, Jobe FW, Pink M, Perry J. An electromyographic analysis of the upper extremity in pitching. J Shoulder Elbow Surg 1: 15–25, 1992.
8. Durall CJ, Manske RC, Davies GJ. Avoiding shoulder injury from resistance training. Strength Cond J 23: 10–18, 2001.
9. Ekstrom RA, Donatelli RA, Soderberg GL. Surface electromyographic analysis of exercises for the trapezius and serratus anterior muscles. J Orthop Sports Phys Ther 33: 247–258, 2003.
10. Ekstrom RA, Donatelli RA, Carp KC. Electromyographic analysis of core, trunk, hip, and thigh muscles during 9 rehabilitation exercises. J Orthop Sports Phys Ther 37: 754–762, 2007.
11. 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.
12. Fenwick CMJ, Brown SHM, McGill SM. Comparison of different rowing exercises: Trunk muscle activation and lumbar spine motion, load, and stiffness. J Strength Cond Res 23: 350–358, 2009.
13. Hislop HJ, Avers D, Brown M. Daniels and Worthingham's Muscle Testing: Techniques of Manual Examination and Performance Testing (9th ed.). St. Louis, MO: Elsevier Saunders, 2014.
14. Jenkins DB. Hollinshead's Functional Anatomy of the Limbs and Back
(9th ed.). St. Louis, MO: Saunders Elsevier, 2009.
15. Kendall FP, McCreary EK, Provance PG, Rodgers MM, Romani WA. Muscles. Testing and Function with Posture and Pain (5th ed.). Baltimore, MD: Lippincott Williams & Wilkins, 2005.
16. Kibler WB, Sciascia AD, Uhl TL, Tambay N, Cunningham T. Electromyographic analysis of specific exercises for scapular control in early phases of shoulder rehabilitation. Am J Sports Med 36: 1789–1798, 2008.
17. McGill SM. Core training: Evidence translating performance and injury prevention. Strength Cond J 32: 33–46, 2010.
18. Neumann DA. Kinesiology of the Musculoskeletal System: Foundations for Rehabilitation (2nd ed.). St. Louis, MO: Mosby Elsevier, 2010.
19. Schoenfield B, Sonmez RGT, Kolber MJ, Contreras B, Harris R, Ozen S. Effect of hand position on EMG activity of the posterior shoulder musculature during a horizontal abduction exercise. J Strength Cond Res 27: 2644–2649, 2010.
20. Smith M, Sparkes V, Busse M, Busse M, Enright S. Upper and lower trapezius muscle activity in subjects with subacromial impingement symptoms: Is there imbalance and can taping change it? Phys Ther Sport 10: 45–50, 2009.
21. Wahl MJ, Behm DG. Not all instability training devices enhance muscle activation in highly resistance-trained individuals. J Strength Cond Res 22: 1360–1370, 2008.
22. Youdas JW, Amundson CL, Cicero KS, Hahn JJ, Harezlak DT, Hollman JH. Surface electromyographic activation patterns and elbow joint motion during a pull-up, chin-up, or perfect pull-up rotational exercise. J Strength Cond Res 24: 3404–3414, 2010.
23. Youdas JW, Arend DB, Exstrom JM, Helmus TJ, Rozeboom JD, Hollman JH. Comparison of muscle activation levels during arm abduction in the plane of the scapula vs. proprioceptive neuromuscular facilitation upper extremity patterns. J Strength Cond Res 26: 1058–1065, 2012.