Competition anxiety and archery accuracy: Cahyo 2026

A hundred and eighty-nine archers who competed between December 2024 and February 2025 in Central Java and the Yogyakarta region filled in the CSAI-2R, the questionnaire measuring competitive state anxiety in its three components, and their scores were taken from official records. They are split by distance: a hundred and three at forty metres, forty-nine at fifty, thirty-seven at seventy.

The three regressions, one per distance, all give a negative coefficient: minus 9.58 points per unit of anxiety at forty metres, minus 10.46 at fifty, minus 14.47 at seventy, with t values between 2.03 and 2.46. The direction is the one everybody expects, and the fact that the coefficient grows with distance is an interesting observation, because an angular error is amplified with distance.

Then comes a second question, whether anxiety changes with distance, and the declared answer is that it changes a great deal, with an F of 44.583: highest at fifty metres, then seventy, then forty.

The problem is that almost none of this article's numbers survives a check, and the most conspicuous case is in the regression table itself. The standardised coefficients reported are minus 1.139, minus 2.041 and minus 2.207. In a regression with one predictor the standardised coefficient is the correlation between the two variables, and a correlation cannot exceed one in absolute value: all three are outside, and two are outside by more than double.

The second question, moreover, cannot be separated from the category. The forty-metre group is aged twelve to nineteen, those at fifty and seventy are aged seventeen to twenty-six. The three distances are three age classes, so saying that anxiety depends on distance is saying it depends on being a youngster or an adult, and the article does not notice this.

A design declared as descriptive quantitative with causal analysis, with anxiety set as the independent variable and cause, and the score as the dependent variable and effect. No intervention: the data come from real competitions and from records.

The sampling is total, that is all the athletes at the competitions, but a hundred and eighty-nine answered voluntarily, and of the non-respondents nothing is known. Where two answers came from the same person the first was kept. University approval, coaches' consent, voluntary participation.

The three groups: forty metres, a hundred and three athletes aged twelve to nineteen, with 7.2 ± 9.4 years of training and 5.9 ± 7.2 of competition; fifty metres, forty-nine athletes aged seventeen to twenty-six, 8.1 ± 6.8 and 6.9 ± 3.6; seventy metres, thirty-seven athletes aged seventeen to twenty-six, 8.2 ± 2.4 and 6.8 ± 0.8.

The instrument is Martens's CSAI-2R in Cox's 2003 revision, with a four-step scale from strong disagreement to strong agreement, and three dimensions: cognitive anxiety, somatic anxiety and self-confidence. There is also an open section for comments, used to enrich the discussion. The questionnaires were distributed before the competition began and collected after it ended: when they were actually filled in is not written, and for a measure of state anxiety that is a piece of information which changes the meaning of the datum.

The analysis: normality with Kolmogorov-Smirnov because the groups exceed thirty cases, homogeneity with Levene, a simple linear regression for each distance, a one-way analysis of variance on anxiety levels across the three distances and post hoc comparisons with the least significant difference method.

What values we are talking about, though, cannot be known: nowhere in the article do means, standard deviations or ranges of anxiety and score appear for any of the three groups. The tables contain the normality tests, Levene's test, the regression coefficients, the analysis of variance sums of squares and the mean differences of the post hoc comparisons. The means themselves are absent, and so is the scale on which the anxiety score is computed, whether a sum or a mean of items.

The gravest fault is in the regression table, and it is one of those visible without redoing any calculation. The standardised coefficients reported are minus 1.139 at forty metres, minus 2.041 at fifty and minus 2.207 at seventy. With a single predictor, the standardised coefficient coincides with the correlation coefficient between the two variables, which by construction lies between minus one and plus one: none of the three can exist. The other three columns of the same table, by contrast, are coherent with each other, because the coefficient divided by its standard error gives exactly the published t value in all three cases. So it is not a widespread error of calculation: it is one column that does not come from where it should.

The p values do not follow from the t values. With a hundred and three cases and one predictor the degrees of freedom are a hundred and one, and a t of 2.031 gives a two-tailed probability of 0.045 and a one-tailed one of 0.022, while the article publishes 0.026, which matches neither. At fifty metres the calculation gives 0.0175 against the published 0.016 and at seventy 0.0207 against 0.019: those two differences are small and could come from rounding, the first one cannot.

One standard deviation is impossible. In the forty-metre group, aged twelve to nineteen, training experience is given as 7.2 ± 9.4 years. Experience cannot be negative and cannot exceed age: with values between zero and nineteen and a mean of 7.2, the arithmetically largest possible standard deviation is 9.22, and setting the realistic limit of thirteen years, that is starting at six, it falls to 6.46. The published value sits above both.

Levene's test has the degrees of freedom of four groups. It reports df1 equal to 3 and df2 equal to 186: with three groups and a hundred and eighty-nine athletes they would be 2 and 186, while a df1 of 3 corresponds to four groups, and with df2 equal to 186 the total would be a hundred and ninety. And there is a reading problem: from Levene's result the article concludes that the samples from the three distances come from the same population, while that test compares variances and not means. The following paragraph finds, with the analysis of variance, that mean anxiety differs a great deal between those same three distances.

The normality tests do not add up either. For anxiety at forty metres the Kolmogorov-Smirnov statistic is 0.192 with a hundred and three cases and a declared probability of 0.071, while at that sample size the critical value at five per cent is around 0.134, falling to 0.087 with the correction used when the mean and variance are estimated from the data: a statistic of 0.192 should give a very small probability, not 0.071. That same 0.192 with that same 0.071 also appears on the row for the score at fifty metres, where with forty-nine cases the value is plausible instead.

The most important limit of substance, though, is not arithmetical, and it is the article's second conclusion. The forty-metre group is aged twelve to nineteen, the other two seventeen to twenty-six: the three distances are also three age classes and three competition categories. Any difference in anxiety between those groups is a difference between youngsters and adults, between different categories, at different competitions, and cannot be attributed to the distance. The sentence by which competition conditions at fifty metres produce more anxiety than at the other distances is not sustainable with this design.

The direction of the link remains open. The questionnaire was handed out before the competition and collected after, and when it was filled in is not stated: the CSAI-2R measures state anxiety, something short-lived that is read at a precise moment, and if an archer filled it in after shooting badly, the link between anxiety and score may arise in the opposite direction from the one the article calls cause and effect. The design is cross-sectional; the article calls it causal as a matter of wording.

Finally the descriptive statistics are missing. No mean, no standard deviation, no range for anxiety and for score in any of the three groups, and the scale of the anxiety score is not declared. Without those, coefficients of about ten points per unit cannot be interpreted, because there is no knowing what a unit is worth, and the post hoc mean differences, 1.043, 1.789 and 2.832, cannot be placed on any scale.

One direction and nothing else can be taken from this article: at all three distances the more anxious archers scored lower. It is what anyone who has shot in competition expects, and here it has the shape of an association measured on a hundred and eighty-nine people with a recognised instrument and with scores taken from official records rather than self-declared, which is the work's merit.

Everything else should be left where it is. The coefficients cannot be used as quantities, because the column of standardised values contains numbers that cannot exist and because the means and the scales are not published: saying that one unit of anxiety costs ten points means nothing if there is no knowing how big a unit is.

The observation that the coefficient grows with distance, from minus 9.58 at forty metres to minus 14.47 at seventy, is the most suggestive thing in the article, because an angular error widens with distance and exactly this is what one would expect. But the three groups are different people, of different ages, at different competitions, so comparing the three numbers is not comparing distances: seeing it would need the same archers shooting at the three distances.

The same holds, with more force, for the second conclusion. That anxiety is highest at fifty metres cannot be taken home: at forty metres the shooters are youngsters aged twelve to nineteen, at the other two distances adults up to twenty-six. What the article measures as an effect of distance is, in good part, the difference between a youth category and a senior one.

One practical thing can be taken from the method, though, and it is a caution for anyone wanting to measure their own archers' anxiety. The CSAI-2R is a state measure: it should be administered at a defined moment, close to the competition and before shooting. Here the questionnaires were handed out before and collected after, and of when they were filled in nothing is known: that alone makes the direction of the link between anxiety and result uncertain. Anyone using it should note the time.