Jackie Hudson, Scholar

---

Measurement/Statistics Articles: Abstract and Introductions

Because traditional procedures of evaluating elastic-like behaviour have yielded mixed results, the purpose of this work was to explore two methods of measuring elastic-like behaviour in the power squat. The intire concentric time method was based on traditional procedures. The initial concentric time method was developed to examine elastic-like behaviour for the beginning 0.2 s of concentric movement. The present study compares a power squat performed maximally by nine subjects at 70% of their 1 repetition maximum. Squats were performed with rebound (REB) and without rebound (NRB). For the entire concentric time method only concentric time was significantly greater (p < 0.05) in the NRB than the REB. For the initial concentric time method the relative displacement, velocity, net work, and peak power of the center of mass were significantly greater (p < 0.05) in the REB than the NRB. Some subjects had theoretically infeasible negative results for elastic enhancement in the entire concentric time method, but not the initial concentric time method. It seems that measuring elastic-like behaviour near the end of the movement can be confounded by the constraints of the task. Based on its success, the initial concentric time method appears to be more appropriate for measurement of elastic-like behaviour in lifting.

In many biomechanical studies it is desirable to know the position or velocity of a round object. When only the product element of velocity is of interest, it is appropriate to use photoelectric devices (Nelson, Larson, Crawford, & Brose, 1966; Roberts, 1972). However, when process elements (e.g., segmental kinematics) as well as product elements are being investigated, the predominant research tools are cinematography and videography.

The biomechanics literature contains many studies in which a cine/video analysis was performed to obtain the position or velocity of a sports implement. Examples of round objects which have been filmed include: baseball (McIntyre & Pfautsch, 1982), basketball (Hudson, 1982), bowling ball (Murase, Miyashita, Matsui, Mizutani, & Wakita, 1974), field hockey ball (House & Owen, 1984), golf ball (Cooper, Bates, Bedi, & Scheuchenzuber, 1974), hammer body (Dapena, 1985), handball (Holt, 1969), lacrosse ball (Stevenson, 1983), netball (Elliott & Smith, 1983), racquetball (Kent & Barlow, 1982), shot put (Dessureault, 1974), soccer ball (Too & Hoshizaki, 1984), softball (Zollinger, 1973), tennis ball (Putnam, 1984), volleyball (Brue & Shapiro, 1981), water polo ball (Davis & Blanksby, 1977), and weight plate of a barbell (Burdett, 1982).

In general, the typical procedure for cine/video data reduction is to: (a) project the film onto a computerized graphic tablet (Owen & Adrian, 1974) or television monitor (e.g., Peak Performance Technologies, 1988), (b) use an attached stylus or cursor to digitize displacement data from previously marked locations, (c) remove random digitizing error in the raw data points by fitting the points with a spline function (Zernicke, Caldwell, & Roberts, 1976) or by treating the points with a digital filter (Winter, Sidwell, & Hobson, 1974), and (d) obtain derivatives by employing direct differentiation of the spline functions or by using finite difference methods (Miller & Nelson, 1973) with the filtered data (Pezzack, Norman, & Winter, 1977). Unfortunately, many conventional aspects of cine/video data reduction are inappropriate when analyzing the behavior of a ball.

First, problems arise at the stage of digitizing because no previously marked location remains at the center of a ball in flight. Although it may be possible to treat a small ball (e.g., golf ball) as a point and estimate its center (Shapiro, 1978), it is doubtful that a large ball (e.g., basketball) could be treated accurately as a point (Disch & Hudson, 1981). Rather than estimating the center of a ball, an alternate procedure is to digitize the top, front, bottom, or back edge of the ball. Given the precision of modern digitizing systems, there are many points which appear to be the topmost, frontmost, etc.. In the best case, the error associated with this method may not be much greater than that associated with digitizing small, round joint markers. However, in the worst case, accuracy of data reduction is diminished if the selected edge is blurred, deformed, obscured, indistinguishable from the background, or leaves the film image.

Second, errors introduced in digitizing are difficult to eliminate. For splines and filters to work properly with round object data, several frames of data are required. Wold (1974) has advised that if there are fewer than 25 data points (frames), the spline method should be used with caution. Although there is no established minimum number of frames to use with a digital filter, the inclusion of additional frames, which bracket the region of interest and are subsequently deleted, is considered necessary (Patrick, Widule, & Hillberry, 1981). If there are too few frames to use splines and filters, it is still possible to fit the data with a polynomial equation, but this method is considered to be less satisfactory than splines (Zernicke, et al., 1976) and filters (Pezzack, et al., 1977). Typically, a biomechanical film of a ball skill might include only 5-10 frames after impact or release. Of course, a larger focal length lens could be used to keep the object in view in more frames, but the compromise is that the subject becomes smaller and more difficult to digitize accurately.

Third, if the digitizing errors are not removed, the inaccuracies are magnified in each subsequent derivative. Although derivatives can be obtained by using finite difference procedures, even when there are a small number of frames with the ball in view, this technique tends to amplify error if the data have not been filtered (Pezzack, et al., 1977).

Because it is imperative for the digitized displacement data of the ball to be accurate so that derived velocity data are valid, a procedure has been developed for estimating the center of a round object from cine/video records. Suggestions for using this procedure to establish reliability and validity in tracking a round object are included also.

In biomechanical research it is common to analyze an activity as it is reproduced in a laboratory or field setting rather than as it is performed naturally at the arena, court, or playground. Despite the greater realism afforded by the natural milieu (where data collection is adjunctive to performance), the contrived setting (where data collection is the objective of performance) is favored because it allows greater control of experimental, extraneous, and error variance. While the control of experimental and extraneous variance is specific to each research design, minimization of error variance is accomplished in biomechanical studies by two general methods: (a) undesirable, extraneous factors (e.g., the distractions of the surrounding environment and people who are superfluous to data collection) are reduced or controlled and (b) precision and accuracy of data collection and reduction are enhanced by the addition of accouterments of biomechanical measurement (e.g., reference scales and axes, cameras, lights, joint markers, light-emitting diodes, electrodes, electrogoniometers, accelerometers, and dynamometers).

Although the use of biomechanical tools adds to the reliability of measurement, using these devices may compromise the external validity of the study if there is a concomitant alteration of performance (Disch & Hudson, 1981). The relationship between alteration of performance and invasiveness of measurement systems has been discussed in the literature. It is widely acknowledged that the more invasive measurement systems, such as electromyography and electrogoniometry, may hinder performance (Adrian, 1971; Miller & Nelson, 1973; Roberts, 1971; Smith, 1976). Although a force plate is minimally invasive, the circumstances of data collection may necessitate the constraint of locomotor performance (Paul, 1976). For example, stride length can range from 56 cm in running two-year-olds (Fortney,1983) to about 4 m in sprinting adults (Bates & Haven, 1973; Deshon & Nelson, 1964); thus, a subject may alter normal stride length to insure that a footfall coincides with a force plate which is approximately 50 cm long. Cinematography is considered to be the biomechanical measurement system which has the least interference with the subject because of the remote position of the camera(s) (Miller & Nelson, 1973; Mitchelson, 1976; Smith, 1976; Whiting, 1976). Nevertheless, it is the opinion of Roberts (1971) that cinematography can have an inhibiting effect on performance.

The purpose of this study was to examine the effect of biomechanical measurement systems on performance. The design of the study was complicated by the fact that it is impossible to quantify the change in performance which is due to the absense of measurement systems without having a measurement system present. Rather than use precise, but unobtainable, variables relating to the process of movement, it was necessary to use a gross measure of the outcome of movement as the dependent variable. It was believed that change in process variables could be inferred from outcome variables by making the following assumption: a given process of movement produces a given outcome for that movement, and, if the outcome is changed, so must be the process. The hypothesis of this study was that performance would diminish with the presence of a biomechanical measurement system.

In the process of conducting research in biomechanics many things can happen that can reduce the reliability and validity of the results. Problems can occur during data collection, reduction and analysis. At the time of data collection errors can be introduced by the measurement systems, the subject, and the environment. Some of the sources of error in cinematography are related to failure to control for errors in parallax and perspective, distortion introduced by the optical system of the camera, selecting a suitable image size, and maintaining the optical axis of the camera perpendicular to the plane of motion. The natural frequency of a force plate or the calibration of a force plate or electrogoniometer can also cause problems during data collection. Timing devices must be calibrated and, if more than one recorder is used, a method of synchronization must be employed. Most errors due to the measurement systems can be minimized by careful control of the experimental procedures.