Fitness clusters in archers: Pratama 2025 study

Eight archers, very likely Indonesian (the paper does not say so explicitly), took four fitness tests and an accuracy test, and their results were sorted into three groups by cluster analysis. Seven cases entered the analysis and the eighth is missing: the clusters hold 4, 2 and 1 archer. The paper prints the analysis of variance that comes with the clusters, and there only two measures separate the groups: flexibility (F 10.107, p 0.027) and core (F 32.334, p 0.003). Arm endurance (p 0.269) and accuracy (p 0.384) do not, and the variable labelled performance has zero variance, so it plays no part in the calculation. The sentence under the table says every test distinguishes the clusters, which the table's own numbers contradict.

The authors do not describe what characterises each group: there are no cluster means, and the example in the discussion (one cluster high in cardiovascular endurance and low in strength, another the reverse) is hypothetical, since neither cardiovascular endurance nor strength was measured.

The methods section is two sentences long and says almost nothing: we are not told who the archers were (age, sex, level, type of bow), where or when they were tested, or how. The variable names in the table point to five measures: flexibility, arm muscular endurance, core, accuracy and a general performance score. The test used for each, the units, and the distance and number of arrows for the accuracy test are not given. Balance appears among the keywords but not in the results.

The analysis sorts the archers into three clusters (the layout of the tables is that of SPSS k-means, though this is inferred from the output) and is followed by an analysis of variance on the same variables; for each archer the distance from the centre of their cluster is reported.

With seven archers in three groups, one of them a single person, there is no basis for talking about profiles. There is also a problem of method that the authors do not mention: clusters are built precisely to make the groups differ on those variables, so an analysis of variance run afterwards on the same variables finds differences almost by construction, and its p values do not carry their usual meaning. Even so, two of the four measures show no difference.

The introduction discusses psychological counselling and a teaching model called CORE, which has nothing to do with core in the sense of trunk muscles; the conclusion is about infrastructure and sports management in Indonesia, which the study did not examine. The journal is flagged in the registry as having questionable editorial standards, and it is unclear whether the paper was peer reviewed.

For anyone who shoots, the study changes nothing. It does not say which physical qualities matter for accuracy, because accuracy does not separate the groups and is not related to the other measures; it does not say how to train an archer, because there was no training; and it does not identify types of archer, because the groups are three clusters of seven people and are not described. The advice in the discussion (strength work for those with high endurance, cardiovascular work for those with high strength) does not come from the data. In the registry it stands as an example of how little a cluster analysis can yield on a sample this small and this poorly described.