Fitness and mental tests to spot archers: Musa 2018

The chapter tries to tell high-potential from low-potential archers by combining fitness tests with a questionnaire on coping with stress, the Athletic Coping Skills Inventory (ACSI). There are five fitness measures: grip strength, arm strength, core strength, static balance and vertical jump. A k-means algorithm run on these measures, the questionnaire score and the shooting score split the archers into two groups. According to the authors, the high-potential group did better on everything except static balance, which did not separate the groups.

Six versions of a k-NN classifier then tried to recognise the two groups. The best, the cosine version, reached 82% accuracy, with a sensitivity of 90.32% and a specificity of 68.42%. The fine and weighted versions reached 80%, the medium and cubic 70%, and the coarse version 62%, because it put everyone in the high group. The values in the table imply that there were 50 archers, 31 in the high group and 19 in the low one, although the chapter does not say so.

The chapter is short and gives few details. The fitness tests follow, it says, the standard procedures described in an earlier paper by the same group. The ACSI questionnaire covers areas such as concentration, coachability and mental preparation, along with how free the athlete is from worry. Tests and questionnaire were collected before the shooting test, for which neither the distance nor the number of arrows is given.

The number of groups was chosen with the silhouette score, computed in Orange 3.11, which pointed to two. The k-NN classifiers (fine, medium, coarse, cosine, cubic and weighted) were run in MATLAB with five-fold cross-validation and assessed on accuracy, sensitivity, specificity, precision, error rate and the Matthews coefficient.

The basic information is missing: the chapter does not say how many archers there were, their age or level, what bows they used, or how the score was obtained. The figure of 50 comes from the table, and the fitness tests are almost identical to those in the same authors' paper on 50 young Malaysian archers: the data are probably the same, but the chapter does not say so.

The design also shares that paper's flaw. The "high" and "low" groups come from an analysis that includes the same measures later used to recognise them, so the accuracy mainly reflects the internal consistency of the method, not an ability to predict who will become a good archer, and nobody was followed over time. The conclusions about the weight of psychological preparation rest on box plots with no statistical tests, and the chapter does not say how much the questionnaire counts compared with the fitness tests. It is a book chapter, and it is not clear whether it was peer reviewed.

For anyone coaching a youth squad, the chapter adds no usable tools. It suggests that among these young archers those who were stronger, jumped higher and scored higher on the coping questionnaire tended to fall in the group with better shooting scores, but it puts no number on any of these links. Looking at body and mind together is a reasonable idea; here, though, it remains an illustrated idea rather than a finding.

The result on static balance, which did not separate the groups, is curious, since balance is often treated as important in selection batteries; with these methods and this sample, however, it supports no conclusion. Accuracy is lower than in the paper using fitness tests alone, 82% against 97.5%, which suggests that figures of this kind depend heavily on choices made in the analysis.