Bands or planks for archers: Purnama 2025 study

Twelve archers from the Blora club in Central Java, between thirteen and seventeen years old: all the registered members, not a sample. Six did sixteen sessions of resistance band work over two months, twice a week, two or three sets of twelve to twenty repetitions at a declared light to moderate intensity. Six did the plank, holds of twenty to thirty seconds, two or three sets, with one to one and a half minutes of recovery. The outcome is a single number, the seconds of hold measured before and after with what the article calls the Holding Bow Test.

With the bands the hold goes from 31.83 ± 4.71 to 41.17 ± 4.45 seconds, up 9.34. With the plank from 32.17 ± 4.62 to 45.67 ± 4.32, up 13.50. The two groups started out practically level, thirty-two seconds each, because allocation paired the baseline scores. Both improve, and redoing the calculations on the published values both improve clearly: the paired test gives t equal to 9.97 for the bands and 20.65 for the plank, on five degrees of freedom.

From there the article concludes that the plank is better. The conclusion is plausible, but the number it rests on does not measure what it is asked to: the abstract writes that the plank showed greater improvement "with correlation values of 0.938 and 0.876". Those two figures are the correlations between the baseline and the final measurement inside each group, that is the line a statistics package prints next to the paired test, and a correlation does not change if the same quantity is added to or taken off every final value. By construction it cannot say how much improvement there was.

The between-group comparison, moreover, is declared significant on the final values, and on the published final values it is not: 41.17 against 45.67 with those standard deviations and six people per group give t equal to 1.777 on ten degrees of freedom, that is p equal to 0.106. The comparison that does hold is a different one, on the gains, which the article does not carry out: 9.34 against 13.50 seconds give p equal to 0.0055. The conclusion survives, for a reason other than the one written.

No arrow, no score, no group training as usual with nothing added.

A two-group design with a before and after measurement and no control. The work took place at the Blora club, in Blora Regency, between August and September 2024, across sixteen meetings at a frequency of twice a week. The twelve athletes were taken in full, that is the whole club, and split into two groups of six with the ordinal pairing technique: baseline scores are ranked and dealt out alternately, so the two groups start out equivalent. It worked, since the two baseline values are 31.83 and 32.17 seconds.

The band group works at light to moderate intensity, two or three sets, twelve to twenty repetitions, one to one and a half minutes between sets. The plank group holds the position for twenty to thirty seconds, two or three sets, with sixty to ninety seconds of recovery. Which band exercises, at what resistance, in which planes of movement: not written. Whether the load was increased over the sixteen sessions: not written. Whether the youngsters went on shooting alongside this, and how much, is not written either, although the introduction says the club works mostly on technique on the field.

The outcome, the Holding Bow Test, is named and not described. There is no definition, no reference to whoever proposed it, no protocol: one cannot read whether the bow is held at full draw or merely held out, at what draw weight, with the archer's own bow or a standard one, by what criterion the stopwatch stops, nor how many trials are run and which is kept. There is no reliability figure. Thirty-two seconds of hold is a high value if the draw is complete and a modest one if the bow is only being supported, and the difference between those two readings changes the meaning of the whole study.

The analysis is declared as follows: normality test, homogeneity test, paired t-test within groups and independent t-test on the final values between groups, threshold 0.05. None of those tests' results are reported beyond the words p less than 0.05: there are no t values, degrees of freedom, exact probabilities, confidence intervals or effect sizes. The results table holds four means, four standard deviations and two differences. Every calculation redone in this card starts from there and from the two correlations quoted in the abstract.

The first limit concerns the number with which the article supports its own conclusion. The abstract says the plank gave greater improvement than the bands "with correlation values of 0.938 and 0.876": a correlation between the baseline and the final measurement says whether the order among athletes was preserved, and stays identical if the same number of seconds is added to every final value. So it measures no improvement, not this one and not anybody's. That these are indeed the paired correlations shows up in a check: using them to recover the standard deviation of the differences, 2.295 seconds for the bands and 1.602 for the plank, the paired tests come out significant as the article declares.

The second is the comparison between groups, and here it does not add up. The article writes that it ran the independent t-test on the final values and found a significant difference in favour of the plank. With 41.17 ± 4.45 and 45.67 ± 4.32 on six and six, that test gives t equal to 1.777 on ten degrees of freedom and p equal to 0.106, both with a pooled variance and with the Welch correction. It is not significant, and it takes nothing to see it: four and a half seconds of difference with standard deviations around four and a half and six people per group.

What is interesting is that the conclusion holds anyway, on a test the article did not run. Since the two groups started out paired, the right comparison is on the gains: 9.34 ± 2.295 against 13.50 ± 1.602 give t equal to 3.641 and p equal to 0.0055. The plank beats the bands on these data, but the route to saying so is not the one written, and the reconstruction rests on the two correlations, so a check against the raw data remains the clean way.

One consequence is worth looking at, though. Those standard deviations of the differences are small: with the plank, six youngsters between thirteen and seventeen would all have gained between roughly eleven and sixteen seconds. A response that uniform, on a voluntary endurance hold and at an age of rapid growth, is unusual.

The third limit is the design. There is no group carrying on training as before, so neither curve can be separated from two months of growth, two months of shooting at the club and familiarity with the test, which was run twice here with no familiarisation session. The comparison between plank and bands is protected from these effects, since both groups undergo them; the claim that each of the two improves the hold is not, and the title and the conclusions make it.

The fourth is the measurement. An endurance hold to exhaustion is a voluntary performance: it depends on how long the athlete decides to keep going, and with twelve youngsters who knew they were in the study, in a small club, with those measuring aware of the groups, nobody was blinded. No arrow was shot and no score was collected, so the chain running from seconds of hold to the target is not even touched.

The fifth concerns the citations, and can be checked against the article's own reference list. That conditioning at that age should use light to moderate loads is attributed to a study on social media use, eating behaviour and nutritional status among adolescents in Jakarta. That the plank requires simultaneous activation of the arms, shoulders and back is attributed to a paper on the relation between physical activity, intensity of social media use and nutritional status in the pupils of one secondary school. That performance in archery depends on keeping the bow position stable is attributed, among others, to an article on integrating deep learning into the primary school curriculum and to one on internalising prophetic values through the practice of archery. That bands increase arm endurance in precision sports is attributed to a survey of the biomotor profiles of elite athletes. Six references out of about twenty sit under statements they do not treat, and the titles are in the article's own list.

What is usable in this article is the dose. Sixteen sessions in two months, twice a week, with bands at two or three sets of twelve to twenty repetitions or planks of twenty to thirty seconds for two or three sets, added to shooting: it is a light load, within reach of a club, and in eight weeks the holding time rises by nine seconds with the bands and thirteen and a half with the plank. On a base of thirty-two seconds that is a third and two fifths more.

The comparison between the two methods is the only thing this design protects, because the two groups started out paired and went through the same two months. Redone on the gains it comes out in favour of the plank, with p equal to 0.0055. If the question is which of the two to put in the programme when there is time for only one, the answer these data support is the plank, and the article's explanation makes sense: an isometric hold resembles the holding phase more than a set of repetitions with a band does.

Two things not to take home. The first: that bands or planks make the hold grow cannot be said from here, because there is no group training as usual, and twelve youngsters between thirteen and seventeen grow on their own over two months. The second: none of this speaks about the score. The hold is a prerequisite, not a result, and an archer who holds the bow ten seconds longer does not put the arrow nearer the centre for that reason.

A practical note on using it in the gym. The article does not say which band exercises, at what resistance, nor whether the load was increased during the sixteen sessions, and it does not describe the endurance test it uses as its outcome. Anyone wanting to replicate the programme has to rebuild it themselves, and anyone wanting to use the same test to measure their own archers has to define it first: full draw or bow held out, with which bow, by what criterion the stopwatch stops. Different choices give numbers that do not compare.