Predicting Athlete Preparation and Performance: A Theoretical Perspective
Sands, Wm A., McNeal, Jeni R., Journal of Sport Behavior
Predicting future events based on knowledge of initial conditions is a basis of the natural sciences. The discovery of covering laws, based on theory, that fully define these predictions is an objective of science. Sport science has suffered from a paucity of efforts to develop covering law theories that frame sport science research limitations and potential benefits. An adaptation of the contingent-necessity model of historical sequences provides a useful, defensible, and testable theory of the development and performance of athletes in real world settings. This model shows that at the contingency-level, precise predictability is extremely limited, while at the necessity- or constraint-level predictability is much more precise.
If a goal of science is the ability to predict future events based on knowledge of initial conditions (i.e., covering laws), then what has sport science been doing in this area for these past decades, and what does sport science have to show for the effort? If we may use a geological metaphor, have sport scientists been simply rock collecting (making observations) instead of developing general theories of historical geology (covering laws) and then finding ways to test these theories (Shermer, 1993). Moreover, has a lack of theory building and testing been due to a lack of data, or have sport scientists basically abandoned the determination of covering laws (i.e., theories) of athlete preparation and performance because the problems are simply too complex?
Data are generally interpreted through, and are evidence for or against a theory (Shermer, 1993). The "data" may be in the form of measurements from experiments or constructed in a narrative form such as historical data. Sport and exercise scientists collect data in order to determine relationships among different variables, but there has been a paucity of attempts to "unify" these data into covering laws or theories. If covering laws of training and preparation are determined, these laws would allow prediction of future events based on a current under-standing of the initial conditions of the athlete, the tasks, and the environment. It is somewhat astonishing that predictability of future athlete performance from currently measurable performance parameters is generally moderate to poor in spite of considerable rhetoric regarding the importance of sport science in the progress of international competition and the viability of elite competitors (Bagiatis, Kellis, & Saroglakis, 1993; Croucher, 1984; Daus, Wils on, & Free-man, 1989; Frederick, 1986; Henry & Stickland, 1994; Jackson et al., 1978; Leonard & Reyman, 1988; Londeree, 1990; Matsudo, 1996; Mcfarlane, 1991; McGarry & Franks, 1994; Meeuwisse, 1991; Mercier, Leger, & Desjardins, 1986; Noakes, 1988; Piper, Ward, McGinnis, & Milner, 1987; Sands & Henschen, 1992; Shephard, 1980; Shermer, 1996; Simpson & Pauson, 1986). According to MacDougall and Wenger: "Laboratory testing should be considered primarily a training aid, not a general tool for predicting future gold medalists. It has severe limitations for identifying potential talent in that scientists still do not know how to determine 'genetic limits' and therefore cannot predict the degree to which an athlete has the potential to improve. Limitations may also exist as to the sport scientist's ability to simulate in the laboratory the physiological demands of some sports. In these cases, as will be reinforced often throughout this book, test results are of little practical value." (Duncan MacDougall & Wenger, 1 991, p3).
Although sport science is touted as a major adjunct to the preparation of elite athletes, sport science is moderate to poor at predicting the performance of any single athlete, even just a few days in the future. Notable attempts have been made at modeling dose-response relationships of athlete training via an operationally defined training impulse. The training impulse is designed to predict athlete performance based on training dosage primarily calculated from heart rate and time measures (Banister, 1991; Banister & Calvert, 1980; Banister, Good, Holman, & Hamilton, 1986; Calvert, Banister, Savage, & Bach, 1976; Fitz-Clarke, Morton, & Banister, 1991). …