MATERIALS AND METHODS
Subjects for this study consisted of 10 healthy male (mean age = 24.2 ± 2.0 yr, mean height = 174.0 ± 5.5 cm, mean mass = 78.7 ± 10.6 kg) and 10 healthy female volunteers (mean age = 21.7 ± 1.5 yr, mean height = 166.1 ± 5.0 cm, mean mass = 66.4 ± 12.9 kg). Based on previous investigations (18,21,22) and the within-subject main effect for contractions intensity on the perceived exertion response, pooled across gender, a sample of six is determined necessary to demonstrate significant effects (1−β = 0.80, α = 0.05, η2 = 0.83). All subjects reported to be physically active, as they indicated performing various types of routine exercise (i.e., jogging, cycling, or use of exercise machines) at least 2 d·wk−1 but were unaccustomed to the prone, single-leg knee flexion contractions. Individuals with a reported history of cardiovascular disease, hypertension, or orthopedic pathology were excluded from participating in this study. The health status of all subjects was ascertained with the Physical Activity Readiness Questionnaire (1). Before all testing, subjects were asked to refrain from engaging in exercise activities at least 24 h before the scheduled testing session. All subjects provided written informed consent as approved by the Institutional Review Board at Eastern Washington University.
Before the testing procedures, the subjects completed a warm-up period that consisted of submaximal cycling on a Monark ergometer for 5 min. The subjects were instructed to self-select a cycling resistance and pedal cadence that felt comfortable to them, and would not induce fatigue. Subjects then performed three unilateral, 20-s hamstring stretches with 20-s rest between stretches. The hamstring stretches were performed to the level of tolerance, as determined by each subject. The subjects were then evaluated for maximal voluntary hamstring muscle contraction (MVC) of their right leg. After performance of the MVC, subjects were assessed for perceived exertion at nine different submaximal intensities each separated by 2-min intervals. After the evaluation session, 10 subjects (5 males and 5 females) were randomly selected and asked to return to the laboratory approximately 1 wk after their initial session to repeat the same procedures. The subjects that were randomly selected all participated in the second session.
Measurement of Isometric Torque
Isometric torque was measured on the Biodex Isokinetic Dynamometer (Biodex Medical, Inc., Shirley, NY). The Biodex Accessory Chair was fully reclined, and subjects lay in a prone position during all testing. The lateral femoral epicondyle was used as the bony landmark for matching the knee joint with the axis of rotation of the dynamometer resistance adapter. Gravity correction was obtained by measuring the torque exerted on the dynamometer resistance adapter with the knee in a relaxed state at full extension. Values for isometric torque were automatically adjusted for gravity by the Biodex Advantage software program (version 4.0). During the measurement of isometric torque, subjects were provided with verbal encouragement as well as visual feedback from the Biodex computer monitor in an attempt to achieve a maximal voluntary effort (10,25). The subjects were specifically instructed to attempt to “push their torque curve to the top of the computer screen.” Subsequent to each MVC, the display window was rescaled such that the previous peak torque was positioned at approximately 80% along the ordinate. For the present study, the MVC was operationally defined to be a maximal voluntary effort produced by each subject with concurrent verbal encouragement and visual feedback of performance. The procedures, including verbal encouragement, were administered to all subjects by the same investigator. Calibration of the Biodex dynamometer was performed according to the manufacturer’s specifications before every testing session.
Once the subjects were comfortably lying prone on the chair, the knee was placed in a position of 30° flexion from full extension, which has been previously shown to be the angle of peak force production for the hamstring muscles during isometric contractions (6,30). Subjects performed two to three submaximal followed by two to three brief maximal-effort isometric contractions for familiarization purposes. Subjects were then asked to perform five MVCs for 5 s each, with at least 2 min of rest in between each MVC. The peak torque of the three highest MVCs was averaged to yield a representative estimate of the individual’s maximal voluntary effort. After approximately 5 min of rest, subjects performed voluntary isometric contractions at the following intensities, in a random order: 10, 20, 30, 40, 50, 60, 70, 80, and 90% of their calculated MVC. The randomization was performed through a sampling without replacement procedure for each subject. This procedure was established before the initiation of the study and involved the random selection of exercise intensities with a table of random numbers. All subjects held each contraction for 5 s with a minimum rest period of 2 min between contractions. Subjects were instructed to match a horizontal line on the Biodex computer monitor with their displayed torque curve that corresponded to the torque level. During all testing, subjects were blinded as to the absolute torque values they were generating. Previous measures of isometric MVC of the hamstring muscles have yielded high test-retest reliability coefficients ranging from 0.81 to 0.99 (30).
Measurement of Perceived Exertion
Perceived exertion was measured with a modified category-ratio scale (CR-10), which eliminated the numerical rating of 0.5 and changed the categorical ratings from “weak” and “strong” to “light” and “hard” (2,17–19). Preliminary work suggested that during isolated muscle contractions, subjects were better able to appreciate these latter descriptors (17). To provide the subjects with a context through which sensation intensities can be evaluated, one high and one low anchor were applied (24). Immediately after each MVC, subjects were instructed to “think about the feelings in your hamstring muscles during the contraction and to assign a rating of maximal to those feelings.” During the recovery period (minimum of 2 min) that proceeded each successive MVC, subjects were instructed to lie quietly and to “think about the feelings in their hamstring muscles and to assign a rating of 0 to those feelings.” In the event that subjects did not report as feeling fully recovered, additional rest time was provided. Immediately after each submaximal contraction (i.e., 10–90% MVC), subjects were instructed to “think about the feelings in their hamstring muscles during the contraction and to rate the feelings of their muscles.” The subjects were instructed to select a number from the modified CR-10 scale and were informed that they may rate a number higher than 10 if they so desired (14). The total number of ratings that exceeded a “10” at each contraction intensity were two at 80% MVC and three at 90% MVC. During the anchoring and all perceived exertion procedures, the subjects observed an enlarged copy (27.9 × 43.2 cm) of the scale.
Hamstring muscle peak torque was expressed as: 1) absolute units (N·m), 2) relative units (N·m·kg−1), and 3) allometric modeled (N·m·kg−n) units and compared between gender. Using the allometric method, the relation between peak torque and body mass were first modeled according to the following power function ratio: y = peak torque·body mass−n, where y is the calculated power function ratio, and n is the exponent of the power function. The natural logarithm (loge) for peak torque and body mass for each subject was calculated and plotted. Linear regression was then applied to estimate the slope (logey = logea + blogeM + cG), where G is gender, M is body mass, and b is the derived exponent. Gender was factored into the regression model as a coded variable (“0” for females, “1” for males) to yield a mass exponent common to both males and females. The y-intercept value generated for males and females were raised to the natural base (e) to determine the proportionality constant (k) of the power function. An independent t-test was performed separately on each measure (i.e., absolute, relative, and allometric modeled MVC) to assess gender differences. An adjustment for an inflated Type I error rate was made for the three t-tests by the Bonferroni-Dunn inequality (alpha level 0.05/3 = 0.017). A two-factor (gender by intensity) ANOVA with repeated measures was performed to assess main and interaction effects for the perceived exertion response at a preset alpha level of P < 0.05. A between-subjects factor for gender and a within-subjects factor for intensity were used in this design.
Perceived exertion was also examined by fitting each subject’s responses across the contraction intensities to the following power function: R = k·Sn, where k is the constant of proportionality, n is any real number, R is the perceived exertion response, and S is the relative contraction intensity. The perceived exertion responses and the relative contraction intensities for each subject were linearized on a log-log plot, via the natural logarithm (loge), and linear regression was applied to each subjects’ responses in order to estimate the slope and the proportionality constant, (logeR = logek + nlogeS). Descriptive data were generated for the calculated exponents and proportionality constants (mean ± standard deviations, and 95% confidence intervals) for males and females, separately. The calculated power function exponent was compared against a value of “1” (unity), to determine nonlinearity, with a one-sample t-test. Gender differences were also examined with an independent t-test. For the purposes of demonstrating the reliability of the present methods, intra-class correlation coefficients (ICC: 2,1) and standard errors of measurement (SEM) were calculated between the three highest MVC.
Between-day reliability was determined by calculating ICC (2,1) and SEM for the following variables: 1) the average of the three highest MVCs on each day; 2) the perceived exertion response at each submaximal intensity (i.e., 10–90% MVC); and 3) the derived exponents and proportionality constants for the perceived exertion measures. The submaximal contraction intensities on day 2 were computed from the established MVC on that day. A two-factor (day by gender) ANOVA with repeated measures was also performed on hamstring muscle MVC between day 1 and day 2. A two-factor (intensity by day) repeated measures ANOVA was performed on the perceived exertion response at a preset alpha level of P < 0.05.
The results demonstrated that males generated significantly greater hamstring muscle torque than the females in absolute (t18 = −5.79, P < 0.017), relative (t18 = −2.88, P < 0.017), and allometric-scaled (t18 = −5.30, P < 0.017) units (Table 1). The derived power functions predicting peak hamstring muscle torque for males and females, respectively, were as follows: peak torque = 17.44·body mass0.357, and peak torque = 12.23·body mass0.357 (r2 = 0.67, SEE = 0.165). The scatterplot demonstrating the relationship between body mass and peak torque is illustrated in Figure 1. The results demonstrated that the perceived exertion response increased significantly across the range of contractions intensities (contraction intensity main effect:F8,144 = 165.01, P < 0.05, η2 = 0.90, 1−β = 0.99). There was no significant gender main effect (F1,18 = 0.43, P = 0.52, η2 = 0.02, 1−β = 0.10) or contraction intensity by gender interaction (F8,144 = 0.93, P = 0.50, η2 = 0.05, 1−β = 0.42) (Table 2). As the results demonstrated no significant gender differences, the perceived exertion data were collapsed across all female and male subjects to derive a power function for the increase in perceived exertion as a function of contraction intensity. The results demonstrated a mean (± standard deviation) power function exponent of 0.89 ± 0.16 (95% C.I. = 0.81–0.96) and a mean proportionality constant of 9.84 ± 1.48 (95% C.I. = 9.15–10.53) (r2 range = 0.89 to 0.99). There were no significant gender differences for the perceived exertion power function exponent (t18 = 0.69, P = 0.50) or the proportionality constant (t18 = −0.085, P = 0.93). The power function exponent was found to be significantly different from a value of “1” (unity), indicating a nonlinear response (t9 = −3.09, P = 0.006).
The peak torque for each of the three highest MVC for all subjects demonstrated high intraclass correlation coefficients (ICC = 0.98, SEM = 4.94 N·m). The mean (± standard deviation) values for each of the three highest MVC were as follows: MVC 1 = 69.55 ± 17.96 N·m, MVC 2 = 69.53 ± 18.62 N·m, and MVC 3 = 69.16 ± 17.08 N·m.
The results for peak torque averaged across the three highest MVCs demonstrated high test-retest reliability between days 1 and 2 (ICC = 0.95, SEM = 2.24 N·m; MVC day 1 = 72.66 ± 20.08 N·m, MVC day 2 = 77.38 ± 22.88 N·m). Although the reliability coefficient for between-day MVC was high, significantly greater values on day 2 than on day 1 were observed (F1,8 = 6.67, P = 0.033, η2 = 0.46, 1−β = 0.62), which was consistent between genders (no significant day by gender interaction:F1,8 = 0.35, P = 0.57, η2 = 0.04). The perceived exertion response demonstrated a significant day main effect (F1,9 = 9.83, P < 0.05, η2 = 0.52, 1−β = 0.80), a significant intensity main effect (F8,72 = 153.98, P < 0.05, η2 = 0.95, 1−β = 0.99), and no significant day by intensity interaction (F8,72 = 0.61, P = 0.77, η2 = 0.06, 1−β = 0.23). In this regard, perceived exertion significantly increased across the contraction intensities on both days, but the overall response was significantly lower on the second day (Fig. 2). The perceived exertion ratings demonstrated low test-retest reliability coefficients across 2 d of testing (Table 3). The derived exponent and proportionality constant (Table 4) also demonstrated low test-retest reliability between the 2 d (exponent: ICC = 0.29, SEM = 0.15; proportionality constant: ICC = 0.18, SEM = 1.82).
The major findings of this study demonstrated that the perceived exertion response during voluntary hamstring muscle contractions at relative submaximal torque levels did not differ between genders, supporting the first hypothesis of this study. This outcome occurred despite the finding that males were able to generate significantly greater hamstring muscle torque than females when corrected for body mass, which confirmed hypothesis 3. The second major finding of the present investigation was the significantly lower perceived exertion ratings on day 2, as compared with day 1. This finding is particularly notable, as the absolute torque levels on day 2 were significantly higher than day 1, which was consistent for both genders. These latter two findings are contrary to what was hypothesized in the present investigation of a repeatable perceived exertion response between days, at the same relative contraction intensities. Given that the reliability portion of the study was comprised of a relatively low sample size (N = 10) and a mixed gender sample, day-to-day repeatability may have been enhanced with more subjects.
The perceived exertion responses during isolated muscle contractions, or resistance-type exercises, have demonstrated a clear dependency on contraction intensity levels (18,21,22) and contraction durations (16,19). With respect to the issue of a gender difference, however, evidence (16) suggests the presence of a gender-specific perceived exertion response during resistance exercise. However, the results of the present investigation supports the more prevalent finding in the scientific literature that healthy young adult females tend to rate their level of perceived exertion during static and dynamic muscle contractions similar to healthy young adult males (18,21,22,27). Contrary to these findings, O’Connor et al. (16) found that young adult females (N = 21) demonstrated higher ratings of perceived exertion during 45 repeated eccentric elbow flexor contractions than healthy young males (N = 21). As the development of muscle fatigue has been shown to be significantly lower in females than males (9), the lower perceived exertion ratings observed by O’Connor et al. (16) may have been reflective of a prolonged endurance capacity for females during isolated muscle contractions. With respect to the present study, however, the influence of relative contraction intensity on ratings of perceived exertion did not appear to be gender specific. The similarity in perceived exertion ratings between females and males also occurred despite the findings of significantly greater torque generating ability in males, when corrected for body mass, than females. Similar results were observed during intensity-incremented quadriceps femoris muscle isometric (18) and dynamic (21,22) contractions.
The generation of isometric knee flexor torque was observed to be significantly greater in males than females, when expressed in absolute, relative, and allometric-modeled units. This finding is consistent with previous investigations examining maximal-effort torque production during knee extension and flexion contractions (6,20). In absolute units (i.e., ft·lb), Fiebert et al. (6) found that males generated significantly greater isometric hamstring muscle torque than females across three different joint angles (30, 60, and 90° flexion). Because the magnitude of muscle force, and hence joint torque production, is positively related to body mass (20), correction of such values to a body dimension has become a standard approach for gender comparisons. A recent investigation demonstrated that males generated significantly greater knee flexor peak torque and work during maximal-effort isokinetic (180°·s−1) contractions, when corrected for body mass (23). However, Neder et al. (13) found that young, middle-age, and older males were stronger than age-matched females for isokinetic knee extensor peak torque in absolute and allometric-scaled but not body mass relative units. Allometric correction of body mass, via power function modeling, demonstrated that gender differences in maximal-effort torque production continued to exist when the torque-body mass relation was included in the scaling procedure. Previous investigations demonstrated power function exponents ranging from 0.63 to 1.054 for various types of resistance exercise (3,13,23,28). These are considerably greater than the exponent value of 0.357 determined in the present study. Inspection of Figure 1 demonstrates relatively greater “scatter” of the data points for females than males, which may explain the negatively accelerating power function fitted to these subjects. Although a separate power function analysis for females and males may help explain the obtained exponent in the present study, the relatively low number of subjects may not yield a truly representative relationship of the examined variables. It is also speculated that this greater force generating capability of males may be reflective of greater cross-sectional areas of existing fiber types (12), and “muscularity” (defined as the ratio of skeletal muscle to adipose tissue-free mass) (29), as compared with females. These differences, however, did not appear to be reflected in the perceived exertion response when the voluntary contractions were performed to similar relative submaximal contraction intensities. At this point, it is not clear how the greater force generating capacity of males, as compared with females, influences the mechanisms driving effort perception. However, the neurological organization of the human male and female, given inherent structural and physiological differences, may be such that the perception of internal and external conditions achieve similarity when stimuli are relatively constant.
An important consideration when evaluating an individuals’ rating of perceived exertion, as well as the ability to generate muscle torque is the variability over successive testing sessions. A major finding of the present study was the significantly lower perceived exertion ratings across all contraction intensities on the second day of testing. It was originally hypothesized that perceived exertion ratings would demonstrate low day-to-day variability, based on reliable estimates of torque output. Although hamstring muscle peak torque displayed a high reliability coefficient (ICC = 0.95) but was observed to be approximately 6.5% greater on the second day, the results suggest that factors not investigated in the present study may have contributed to the perceived exertion response. It is well accepted that numerous factors, such as neuromotor, psychological, and environmental factors, play a role in shaping the perceived exertion response during exercise (14). Robertson and Noble (24) have further suggested that a cognitive reference filter functions as a final step in the determination of a perceived exertion rating, which incorporates previous exposure to the stimulus. This mechanism may have been an important factor in the present investigation as dissociation between voluntary torque output and perceived exertion ratings were apparent across the two testing days. Contrary results were observed by Elfving et al. (5) in 11 healthy adults performing a submaximal, 45-s isometric back extension followed by a subjective rating of fatigue obtained with the Borg CR-10 scale. Performed over three nonconsecutive days, Elfving et al. (5) demonstrated a perceived exertion reliability coefficient of ICC = 0.84 and a 0.90 SEM. It is probable that additional exposure through subsequent testing days may have reduced the between-day difference in the present study, such that habituation to the exercise task may have been achieved. It should be reiterated that one investigator administered the testing procedure, in the same manner, to all subjects. The clinical use of perceived exertion ratings for resistance exercise evaluation and prescription must be predicated on established reliability and expected variability data. Thus far, such data are limited in the scientific literature. Based on the results of the present study, in consideration of the modest sample size, the day-to-day assessment of perceived exertion may not be reliable within subjects.
The results from the present study demonstrated that day-to-day variability of perceived exertion during hamstring muscle contractions was high, and the ratings were consistently lower on a second day of assessment. It may be argued that habituation to an unaccustomed exercise task was largely responsible for such findings, but this remains speculative. Furthermore, the finding that a gender-specific perceived exertion response did not exist, despite significantly different strength values between females and males, is suggestive of a similar stimulus-response relationship as a function of contraction intensity. Although the experimental procedures focused on an isometric task at a single joint angle, as well as generating high day-to-day variability, the findings suggest that a period of familiarization may be warranted when evaluating perceived exertion during voluntary muscle contractions.
1. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription, 6th Ed.,
B. A. Franklin, M. H. Whaley, and E. T. Howley (Eds.). Baltimore: Lippincott Williams & Wilkins, 2000, pp. 22–32.
2. Borg, G. A. V. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 14: 377–381, 1982.
3. Davies, M. J., and G. P. Dalsky. Normalizing strength for body size differences in older adults. Med. Sci. Sports Exerc. 29: 713–717, 1997.
4. Dedering, A., M. R. A. F. Hjelmsater, B. Elfving, K. Harms-Ringdahl, and G. Nemeth. Between-days reliability of subjective and objective assessments of back extensor muscle fatigue in subjects without lower-back pain. J. Electromyogr. Kinesiol. 10: 151–158, 2000.
5. Elfving, B., G. Nemeth, I. Arvidsson, and M. Lamontagne. Reliability of EMG spectral parameters in repeated measurements of back muscle fatigue. J. Electromyogr. Kinesiol. 9: 235–243, 1999.
6. Fiebert, I. M., C. H. Pahl, E. B. Applegate, N. I. Spielholz, and K. Beernik. Medial-lateral hamstring electromyographic activity during maximum isometric knee flexion at different angles. Isokinetics Exerc. Sci. 6: 157–162, 1996.
7. Gwinn, D. E., J. H. Wilckens, E. R. Mcdevitt, G. Ross, and T. C. Kao. The relative incidence of anterior cruciate ligament injury in men and women at the United States Naval Academy. Am. J. Sports Med. 28: 98–102, 2000.
8. Hollinshead, W. H., and C. Rosse. Textbook of Anatomy, 4th Ed. Philadelphia: Harper & Row, 1985, pp. 349–400.
9. Hunter, S. K., and R. M. Enoka. Sex differences in the fatigability of arm muscles depends on absolute force during isometric contractions. J. Appl. Physiol. 91: 2686–2694, 2001.
10. Kim, H. J., and J. F. Kramer. Effectiveness of visual feedback during isokinetic exercise. J. Orthop. Sports Phys. Ther. 26: 318–323, 1997.
11. Legally, K. M., R. J. Robertson, K. I. Gallagher, et al. Perceived exertion, electromyography, and blood lactate during acute bouts of resistance exercise. Med. Sci. Sports Exerc. 34: 552–559, 2002.
12. Miller, A. E., J. D. Macdougall, M. A. Tarnopolsky, and D. G. Sale. Gender differences in strength and muscle fiber characteristics. Eur. J. Appl. Physiol. 66: 254–262, 1993.
13. Neder, J. A., L. E. Nery, A. C. Silva, S. Andreoni, and B. J. Whipp. Maximal aerobic power and leg muscle mass and strength related to age in non-athletic males and females. Eur. J. Appl. Physiol. 79: 522–530, 1999.
14. Noble, B. J., and R. J. Robertson. Perceived Exertion. Champaign IL: Human Kinetics, 1996, pp. 4–71.
15. Norkin, C. C., and P. K. Levangie. Joint Structure and Function: A Comprehensive Analysis, 2nd Ed. Philadelphia: F. A. Davis, 1992, pp. 337–378.
16. O’Connor, P. J., M. S. Poudevigne, and J. D. Pasley. Perceived exertion responses to novel elbow flexor eccentric action in women and men. Med. Sci. Sports Exerc. 34: 862–868, 2002.
17. Pincivero, D. M., S. M. Lephart, N. M. Moyna, R. G. Karunakara, and R. J. Robertson. Neuromuscular activation and RPE in the quadriceps at low and high isometric intensities. Electromyogr. Clin. Neurophysiol. 39: 43–48, 1999.
18. Pincivero, D. M., A. J. Coelho, and W. Erikson. Perceived exertion during isometric quadriceps contractions: a comparison between men and women. J. Sports Med. Phys. Fitness 40: 319–326, 2000.
19. Pincivero, D. M., and W. S. Gear. Quadriceps activation and perceived exertion during a high intensity, steady state contraction to failure. Muscle Nerve 23: 514–520, 2000.
20. Pincivero, D. M., W. S. Gear, R. L. Sterner, and R. G. Karunakara. Gender differences in the relationship between quadriceps work and fatigue during high intensity exercise. J. Strength Cond. Res. 14: 202–206, 2000.
21. Pincivero, D. M., A. J. Coelho, R. M. Campy, Y. Salfetnikov, and A. Bright. The effects of voluntary contraction intensity and gender on perceived exertion during isokinetic quadriceps exercise. Eur. J. Appl. Physiol. 84: 221–226, 2001.
22. Pincivero, D. M., A. J. Coelho, and R. M. Campy. Perceived exertion and maximal quadriceps femoris muscle strength during dynamic knee extension exercise in young adult males and females. Eur. J. Appl. Physiol. 89: 150–156, 2003.
23. Pincivero, D. M., C. B. Gandaio, and Y. Ito. Gender-specific knee extensor torque, flexor torque, and muscle fatigue responses during maximal effort contractions. Eur. J. Appl. Physiol. 89: 134–141, 2003.
24. Robertson, R. J., and B. J. Noble. Perception of physical exertion: methods, mediators, and applications. In: Exercise and Sport Science Reviews, J. O. Holloszy (Ed.). Baltimore: Williams & Wilkins, 1997, pp. 407–452.
25. Sahaly, R., H. Vanderwalle, T. Driss, and H. Monod. Maximal voluntary force and rate of force development in humans: importance of instruction. Eur. J. Appl. Physiol. 85: 345–350, 2001.
26. Solomonow, M., R. Baratta, B. H. Zhou, et al. The synergistic action of the anterior cruciate ligament and thigh muscles in maintaining joint stability. Am. J. Sports Med. 15: 207–213, 1987.
27. Tomporowski, P. D. Men’s and women’s perceptions of effort during progressive-resistance strength training. Percept. Mot. Skills 92: 368–372, 2001.
28. Vanderburgh, P. M., and C. Dooman. Considering body mass differences, who are the world’s strongest women? Med. Sci. Sports Exerc. 32: 197–201, 2000.
29. Wang, Z., M. Heo, R. C. Lee, D. P. Kotler, R. T. Withers, and S. B. Heymsfield. Muscularity in adult humans: proportion of adipose tissue-free body mass as skeletal muscle. Am. J. Hum. Biol. 13: 612–619, 2001.
30. Worrell, T. W., G. Karst, D. Adamczyk, et al. Influence of joint position on electromyographic and torque generation during maximal voluntary isometric contractions of the hamstrings and gluteus maximus muscles. J. Orthop. Sports Phys. Ther. 31: 730–740, 2001.