Training methods and recurve scores: Yamanarda 2025

Thirty recurve archers between fifteen and twenty-four, licensed with an Istanbul club, split into three groups of ten by stratified random allocation on age. One group adds eight weeks of scapular stabilisation exercises, twice a week; another adds eight weeks of scapular proprioceptive neuromuscular facilitation administered by a physiotherapist; the third carries on with its coach's programme alone. The outcome is the score of six arrows at eighteen metres on an indoor face, after four warm-up arrows that do not count.

The three groups start out level on age, years of practice, height and weight, with p values between 0.73 and 0.97, and level on the score too, 46.70, 45.00 and 44.70 points out of sixty, with F equal to 1.509 and p equal to 0.239. After eight weeks: 55.20 with stabilisation, 52.80 with facilitation, 47.40 in the control group. Plus 18.2 per cent, plus 17.3 and plus 6.0. The time by group interaction gives F equal to 70.846, so the groups did not improve in the same way, and between the two kinds of exercise there is no difference.

One thing has to be said about this article straight away, and it cannot be said about the others in the same line: the arithmetic all holds. Recomputed from the published values, the three partial eta squared figures in the repeated measures table come out at 0.969, 0.438 and 0.840, exactly the three published; the three d values are exactly the three t values divided by the root of ten; and the interaction F, rebuilt from the gains the three paired tests imply, comes out at 70.846, fourth digit included. The five tables are one coherent structure.

Precisely because it is coherent, it can be used to recover what is not printed: the standard deviations of the scores, which the main table lacks. They come out at 2.78 points between archers before and 2.12 after, and the individual change between the two measurements has a standard deviation of about 1.19 points. That last number is the problem: it says six arrows measure an archer to within less than a point, and six arrows cannot do that.

An experimental design with three groups and measurements before and after, December 2024 and January 2025, on archers from an Istanbul club training with recurve bows. Entry criteria: aged between fifteen and twenty-four, training regularly, at least a year of practice, no upper limb injury in the past six months, voluntary consent. Thirty participants, ten per group. Allocation is a draw within age strata, and then a second draw decides which group does which protocol: it is an orderly procedure, and the baseline table shows it worked. Ethical approval from Üsküdar University, parental consent for those under eighteen.

Scapular stabilisation is four exercises: scapular retraction lying prone with the elbows extended, a push-up against the wall with a ball, retraction with a resistance band with the arms at ninety degrees of abduction, and the so-called scapular clock, in which with the hand on the ball against the wall one traces a clock face through three, six, nine and twelve o'clock. The facilitation is applied by an experienced physiotherapist on the two patterns of anterior elevation with posterior depression and anterior depression with posterior elevation, with the repeated stretch technique: the agonist muscle is taken to maximum length, then a rapid stretch, then isotonic contractions against resistance through the whole range of movement.

The control group is described precisely, which is not common, and the description matters: it carries on with its coach's programme, which includes one-armed band rows, weighted work for the biceps and triceps, and weighted exercises for the shoulder girdle, deltoid, trapezius and rotator cuff. So the control group is not a group doing no shoulder work: it is a group doing shoulder work its own way. The comparison is between two added protocols and a programme already under way.

The shooting test follows the indoor regulations of the Turkish and the world federation: ten arrows at eighteen metres, the first four for warm-up and discarded, the last six scored, on a concentric face scoring from ten in the inner gold to five in the white. The same test before and after.

The analysis: Shapiro-Wilk for normality, one-way analysis of variance for the between-group comparisons at each moment, the paired t-test within groups, repeated measures analysis of variance for time and for the time by group interaction, and Tukey's test as the post hoc comparison. The threshold is 0.05. The main table reports means, percentage changes, F, eta squared and p, but no standard deviations.

The limit that weighs on everything is the measurement: six arrows. The standard deviations are not printed, but they can be recovered from the F and eta squared values, and they come out at 2.78 points between archers at the first measurement and 2.12 at the second. From the three paired tests one also recovers the standard deviation of the individual changes, which is about 1.19 points. Since the change between two measurements contains the instrument's noise twice, that number implies that a score over six arrows pins an archer down to within 0.84 points, and therefore that a single arrow has a standard deviation of 0.34 points on a scale running from five to ten. An archer averaging 7.8 per arrow with a standard deviation of 0.34 would put nearly every arrow between the seven and the eight, and never a ten or a six. Arrows do not group like that. The d value of 8.75 the article reports for scapular stabilisation comes from here: it is not a large effect, it is a spread too small for the number of arrows shot.

This is not a fault in the arithmetic, it is a fault in the instrument, and there is only one way to settle it, by shooting more arrows. Six arrows are one end. An end's score swings from end to end for reasons that have nothing to do with training, and that is why indoor regulations count sixty arrows.

The second point concerns the post hoc comparison, and here the article answers a different question from the one it asks. The note on the Tukey table says the difference between groups is computed "as the average of pre-test and post-test results", and indeed the three values, 2.05, 4.90 and 2.85, are exactly the differences between the groups' means across the two moments. That is the between-subjects quantity, which mixes the starting point with the gain, while the sentence that follows reads it as a greater improvement. The comparison on the gains the article does not make. Redoing it, the conclusion holds and strengthens: stabilisation against control gives 5.80 points more gain and facilitation against control 5.10, both well past the threshold, while between the two protocols 0.70 points and no difference remain. The published p of 0.034 for facilitation against control therefore understates the difference, because that test dilutes the gain with the starting level.

The third: the scapula was not measured. The intervention is scapular on both arms of the trial and the only outcome is the score. No dyskinesis angle, no strength of the serratus anterior or lower trapezius, no range of movement, neither before nor after. The discussion fills the gap by citing other studies that did measure those things, but in this sample the mechanism remains unverified: it is known that the two groups improved their score, not that they changed their scapula.

The fourth is the dose. The four stabilisation exercises and the two facilitation patterns are described by name, with a reference to whoever proposed them, and with no sets, repetitions, hold times, resistances or progression. Twice a week for eight weeks is all one reads. Anyone wanting to reproduce the protocol has to rebuild it.

The fifth is the comparison with the control group. The two experimental groups received sixteen extra sessions, in a university physiotherapy centre, and the facilitation was applied by hand by a physiotherapist: that carries attention, the journey, the supervised session and the contact, and the control group had none of it. Nobody was blinded, and the outcome is a shooting score, that is a voluntary performance. It should be acknowledged all the same that the control group's programme already included shoulder work, so the comparison is not between working on the scapula and not working on it.

The sixth is the control group itself: plus 2.70 points, 6 per cent, with p below 0.001 and a d of 2.55. A group that changed nothing gained a third of what the best group gained. That part of the gain comes from having repeated the same test, and the same component sits inside the other two groups' numbers.

Finally two smaller things. The sample size calculation says thirty participants suffice for a moderate effect, f equal to 0.30, with eighty per cent power: for a one-way analysis of variance on three groups that effect calls for about a hundred and ten participants, and with thirty the power is around twenty-five per cent. The claim can hold only for the repeated measures interaction with a high correlation between the two measurements, which the article does not state. And the dates do not line up: the article says it derives from a master's thesis accepted in 2022, with ethical approval from October 2022, while the participants are described as archers from the period between December 2024 and January 2025. The participants' sex, collected on the demographic form, is not reported.

This is the best-built work in its line, and it is worth saying why: three groups drawn by lot, a control group described in full, baseline data published and level, protocols with a reference to whoever proposed them, and a statistical apparatus that recomputes to the fourth digit. Two conclusions come out of it, and they hold.

The first: eight weeks of scapular work added to training, twice a week, shows up on the score, and shows up even when the base programme already contains rows and weighted shoulder exercises. Redone on the gains, the advantage over the control group is 5.8 points out of sixty for stabilisation and 5.1 for facilitation, with a control group that gains 2.7 on its own in the meantime.

The second: between the two methods nothing changes. The repeated stretch applied by hand by a physiotherapist and four exercises an archer can do alone with a ball and a band reach the same result, 0.70 points apart. For anyone having to choose, that is the information that counts: the version needing no therapist is just as good, and it can go into a club's programme without depending on anybody.

The four exercises, incidentally, are within anyone's reach: scapular retraction lying prone with the elbows extended, a push-up against the wall with a ball, retraction with a band with the arms open at ninety degrees, and the scapular clock, the hand on the ball against the wall running through three, six, nine and twelve o'clock. What the article does not give is the dose: no sets, no repetitions, no hold times. Two sessions a week for eight weeks is the only reproducible number.

One thing, though, should not be taken home, and that is the size of the effect. The outcome is the score of six arrows, one end, and the data imply that this measurement would have a margin of under a point, which six arrows cannot deliver. So the direction of the result is credible and its magnitude is not: eighteen per cent improvement in eight weeks is a figure not to be repeated. Anyone wanting to test the protocol on their own group should measure over sixty arrows, as in competition, and expect smaller numbers.