Strength, focus and archery: Sulistiyowati 2026

Twenty archers from the Vijas club in Sidoarjo, East Java, measured once on three things: pulling strength on an expanding dynamometer, concentration on the Grid Concentration Test and the score at thirty metres. No training, no intervention: the study only looks at how far the three measures move together.

The correlations are high. Between concentration and score r is 0.616 with p equal to 0.004; between strength and score 0.927 with p below 0.001. In the two-variable regression strength stays and concentration disappears: every unit on the dynamometer is worth 6.46 points, with a standardised beta of 0.998 and p below 0.001, while concentration takes a negative coefficient, minus 1.24, with a beta of minus 0.098 and p equal to 0.453. Hence the article's conclusion, that arm strength is the dominant factor.

The arithmetic of this part is impeccable, and it is worth saying because it is not common. Redoing the regression from the published values alone, the two betas come out, the two t values come out to three decimals (7.81 and minus 0.767), the raw coefficients come out (6.46 and minus 1.24), and so does the constant, 181.55 against the published 181.591. The three tables are coherent with one another.

That very coherence brings out the number the article does not publish. For the betas to be what they are, the correlation between strength and concentration in this group has to be 0.716, and the two equations give it independently of each other, 0.7165 and 0.7160. So in this sample a pull on a piece of gym equipment and a test of finding numbers on a grid agree to seventy-two per cent. Between those two things no mechanism exists: what exists is a third factor moving both.

That third factor shows up in the descriptive data. Strength runs from 5 to 30, the score from 180 to 340, concentration from 11 to 23, and about the sample nothing is known: not age, not sex, not years of practice, not division, although the article says the club has recurve, national and compound archers. Twenty people whose strength varies sixfold, lined up along a score that varies twofold, are a snapshot of a club holding both children and adults.

A correlational design, a single session, April 2026, on the Vijas club's training field. The population is the club's twenty active athletes, registered with the Indonesian federation for East Java, and the sample is those same twenty: the article calls it purposive sampling, but having taken everybody there is no purpose left to serve. The declared criteria are being an active club athlete, being able to shoot at thirty metres, knowing the basic technique and being willing to take part.

Three instruments. Arm strength on an expanding dynamometer, two trials with the better one kept: this is the sprung device pulled apart in front of the chest with both hands, so it measures a two-armed horizontal pull and not the act of drawing a bow, which is asymmetrical and constrained by the geometry of the body. Concentration on the Grid Concentration Test, in which numbers are found in order on a grid within a set time and the count is how many were found. Accuracy as the total score on a regulation face at thirty metres.

How many arrows is not written: we read that each athlete was given the chance to perform "a predetermined number of shots according to the testing procedure", and the procedure does not appear. Since the highest score observed is 340, the arrows must be at least thirty-six, that is a maximum of at least 360, but that is a deduction and not a datum. Also missing are the draw weight, each archer's bow type and division, and any information on wind conditions.

The order of the tests is fixed for everybody: strength first, then concentration, then shooting. That means everybody shot after two maximal pulls on the dynamometer and after a timed cognitive task, and that any effect of that order entered the data the same way for everybody, with no way of separating it out.

The analysis: descriptive statistics, the Shapiro-Wilk normality test because the sample is under fifty, a linearity test, Pearson correlation and multiple linear regression, with the threshold at five per cent. The results of the normality and linearity tests are not reported. The descriptive table labels the three variables as minutes, times and seconds, while they are respectively a count of numbers found, a force on a dynamometer and a score in points: three units of measurement out of three do not match what was measured.

The limit that matters is the interpretation of the highest coefficient. A correlation of 0.927 between a pull on an expanding dynamometer and a score at thirty metres means that pull accounts for eighty-six per cent of the variance in twenty archers' scores. No single physical quality gets there, in any sport, if the sample is homogeneous. It gets there at once, in any sport, if the sample is not: it is enough for the twenty to include youngsters and adults, because then strength measures size and the score measures experience, and the two rise together without either causing the other. That this is the case is suggested by the three declared ranges, strength from 5 to 30 and score from 180 to 340, and confirmed by the fact that not a word is written about age, sex, years of practice or division.

The second point comes from the internal coherence of the regression, and it is the most informative thing in the whole article. The two published betas imply, algebraically and by two independent routes, a correlation of 0.716 between dynamometer strength and the concentration test score. That number is not in the article. Between pulling a spring and finding numbers on a grid there is no plausible physiological link: a correlation like that, in twenty people, is the signature of a common gradient running through the sample. The same gradient also explains why concentration on its own correlates 0.616 with the score and then is worth nothing once strength enters the model. The full model accounts for 86.5 per cent of the variance, strength alone for 85.9: concentration is worth six tenths of a percentage point.

The third is the design, which the article states precisely and then treats as if it were another. It is a survey on one day with no treatment, and the conclusions write that arm strength is the main determinant "in improving" accuracy and that the study shows physical training has to become the primary focus. Twenty measurements taken once will not yield the finding that training one thing improves another, nor even the direction of the link: an experienced archer shoots more and is therefore stronger, or is stronger and therefore shoots better, and these data do not tell the two apart.

The fourth is the instrument. The expanding dynamometer measures a two-armed pull in front of the chest, while drawing a bow is an asymmetrical act in which one shoulder pushes and the other pulls, with the scapula closing behind. Taking the first as a measure of the second is a shortcut, and the article does not discuss it.

The fifth concerns the reference apparatus, and can be checked against the printed list. Seven references out of twenty-two have, in place of an author, something that is not a person: a journal's name ("Teori, Jurnal Psikologi"), a department and a university ("Keolahragaan, S Ilmu, Fakultas Ilmu Olahraga"), a university and a province ("Maret, Universitas Sebelas, and Central Java"), the words of the title itself ("Effect, T H E, O F Crossfit, Training On"), and in one case "Duolingo, Waktu Layar, and Capaian Grammar". That last one is a paper on language learning with Duolingo, and in the methods it is the reference used to justify the choice of the Shapiro-Wilk test below fifty subjects. Two entries carry the words "No Title" as their title, and one of those two is the reference supporting the sentence by which all the instruments were checked and found to comply with measurement standards. One article appears twice, the same issue of the same journal, once as "Anderzen et al. 2026" and once as "Effect et al. 2026". And among the introduction's references are a study on the effect of activity videos on the physical activity of four, five and six-year-olds in preschool and one on plyometric interval training in physical education lessons, brought in to support statements about archery.

The last figure is editorial and sits in the header: received 23 May 2026, reviewed 24 May, accepted 2 June. One day between arrival and review, ten in all.

No training decision comes out of this study, and the reason is not the result: it is the sample. Twenty archers from a club in which measured strength varies sixfold are not a group inside which the link between a physical quality and the score can be read, because that link is produced by the difference in level between the people. The same work, redone inside a single category and with years of practice controlled, would give a much lower number, and that would be the useful one.

There is one thing worth taking away, though, and it comes from comparing the two halves of the analysis. Taken alone, the concentration test correlates with the score at 0.616, which looks like a lot. Put beside strength, it adds nothing: the model with both accounts for 86.5 per cent of the scores, strength alone for 85.9. This is why a single correlation, in a mixed group, does not establish that the quality matters: it may be the reflection of something else that is measured better. Anyone reading studies on scores and physical qualities does well to look always for the second table, the one where the variables sit together.

In practical terms two concrete observations remain. The first is that the expanding dynamometer, the sprung device pulled with both hands in front of the chest, does not measure the draw: the draw is asymmetrical and the useful force is in the closing of the scapula, not in a symmetrical push of the arms. Anyone wanting a number to follow over time would do better to measure draw force on the bow, or the hold at full draw, which at least reproduce the geometry of the movement. The second is that a score at thirty metres with no stated arrow count compares with nothing: if a log is kept, the arrow count and the distance belong next to the score, always.

For the rest, the idea that arm strength is the dominant factor in accuracy is a hypothesis this work does not test. Testing it would mean following the same archers while their strength changes, and seeing whether the score follows.