Competition anxiety in archers: Abdoshahi 2021 study

At the 2011 Asian Championships in Tehran, 57 national-team recurve archers, 28 men and 29 women, rated themselves on two things five to ten minutes before going out to shoot, at three stages of the same competition: qualification, the individual matches and the team event. The two measures are cognitive anxiety, that is how worried they felt, and felt arousal, that is how keyed up the body felt.

Cognitive anxiety changes between stages. Before qualification the mean is 3.67, before the individual matches 5.21 and before the team event 5.05, with an effect of stage of F(2,110) = 15.27, p < 0.001 and partial eta squared 0.22. The two knockout stages therefore sit above qualification, but do not separate from each other: the authors' hypothesis, that the individual matches would be the heaviest stage even compared with the team event, did not show up.

Felt arousal, by contrast, does not move: F(2,110) = 0.13, no difference between the three stages. What rises as the matches approach is the worry, not the sense of being keyed up.

Between men and women the difference is there on both measures, as a main effect: arousal 7.61 against 6.85, F(1,55) = 13.23, p < 0.001, eta squared 0.19; cognitive anxiety 5.61 against 3.64, F(1,55) = 26.63, p < 0.001, eta squared 0.33. There is no interaction between sex and stage, however, either on anxiety (F = 0.22) or on arousal (F = 1.56): the gap between the two groups stays much the same across all three stages. Age, years of practice and number of international competitions did not differ between the groups.

No shooting scores were collected: the study measures how the archers felt, not how they shot.

At the 2011 championship there were 99 men and 72 women from 22 countries. The study chose recurve because it was the largest class, and out of the 113 recurve archers it enrolled 57: the volunteers who would compete in all three stages. They were all members of their national teams, with normal vision and right-handed, and signed written consent. Permission to collect the data was obtained from the organisers, with a review of ethical issues; the coaches had been informed at registration and helped with recruitment.

The instrument is the Sport Grid-Revised, a nine-by-nine grid on which a single point is marked: the horizontal axis runs from «not worried» to «very worried» and measures cognitive anxiety, the vertical axis from «very low activation» to «very high activation» and measures felt arousal. The authors, however, did not use the grid: they broke it into two separate nine-point scales, one per variable. The reasons given are two, the little time available at a competition of that level and the fact that some archers did not have good enough English to understand how to combine two variables into a single point on the grid.

The two questions were put five to ten minutes before the archer left for their event, at each of the three stages, by two investigators with good English; where needed, the team's translator was present.

The analysis is a mixed-design analysis of variance, 2 by 3, with sex between subjects and stage within subjects, Greenhouse-Geisser correction where sphericity did not hold, and post-hoc comparisons with Bonferroni-corrected t tests. Partial eta squared and Cohen's d are reported as effect sizes. The software is SPSS 24.

It is worth knowing how the competition was built, because it changes the reading. In individual recurve at that championship 64 men and 49 women competed, and all of them went through to the matches after the 72 arrows of qualification: qualification only served to set the brackets, it eliminated nobody. The individual matches were in sets of three arrows, with two points for a set won and the first to six points winning; the team matches, with three archers of the same sex per team, were in sets of six arrows, three per team in turn. Fifteen men's teams and eleven women's teams took part in the team event.

Two figures reported as effect sizes do not agree with the means and standard deviations in the table. The authors give a Cohen's d of 1.64 for the difference in arousal between women and men and 1.84 for the one in cognitive anxiety: these are enormous effects. But the arousal difference is 0.74 points on a nine-point scale, with standard deviations between 0.80 and 1.71, and the anxiety difference is 1.93 points with standard deviations between 1.42 and 2.49. Redoing the calculation from the table gives about 0.6 for arousal and about 1.0 for anxiety, that is half the reported values (the recalculation is mine). The same error shows in the degrees of freedom of the post-hoc comparisons, given as t(169): 169 is the 171 observations of the 57 archers across the three stages minus two, that is each archer's three measures were treated as independent. With 57 people the degrees of freedom for a comparison between sexes are 55, as in the main effects of the analysis of variance. The direction of the differences does not change, and the partial eta squared values (0.19 and 0.33) come from the correct model; it is the d values and the t tests that are inflated.

The abstract reads as though it contained a contradiction: it says there are no sex-based differences in competitive anxiety once the stages are considered, and immediately afterwards that the women had more arousal and more anxiety than the men across all stages. The two sentences speak about different things: the first about the interaction between sex and stage, which is absent, the second about the main effect of sex, which is present. The discussion repeats the same formulation, and a reader who takes in only the abstract risks getting the result backwards.

Any measure of performance is missing. There are no scores, no match outcomes, so the study says nothing about what effect anxiety or arousal had on how these archers shot. The practical conclusions, about reducing gender stereotypes and teaching stress management strategies, do not rest on data from this study.

The instrument was modified. The validity and reliability of the Sport Grid-Revised were established for the nine-by-nine grid used as a single point; here it was broken into two separate single-item scales, and no check on that version is reported. The reasons for the split are understandable, but what remains is two self-rated single items, in a language part of the sample did not command, with their own team's translator alongside.

The authors declare four limitations and they are worth repeating: the design is cross-sectional and self-reported, so it does not allow causes to be inferred; other variables that matter were not measured, such as experience at high levels and self-confidence; neither trait anxiety nor a baseline away from competition days was measured, so a starting difference between the sexes cannot be ruled out; and the data were collected ten years before publication.

The sample and the format remain. The 57 archers are volunteers out of 113, hence self-selected, and all from Asian national teams at a single championship. The qualification stage of that competition eliminated nobody, so the comparison is between a stage with no cut and two knockout stages: it says nothing about competitions in which qualification cuts the field. The team stage, moreover, differs not only in pressure but in task, because three archers shoot with a shared responsibility. Finally, the discussion's long explanation in terms of gender stereotypes, stereotype threat and socialisation rests on nothing measured here: none of those variables was recorded. One detail: the methods give the mean age as 26.12 ± 0.34 years, while the table reports 26.40 ± 3.90 for the men and 27.10 ± 5.40 for the women, and the significance threshold is written as «0.05%».

The result that matters to anyone competing is the dissociation between the two measures. As the knockout matches approach the worry rises, from 3.67 to over 5 on a nine-point scale, while the sense of being keyed up stays identical, around 7, across all three stages. It is not the body changing gear between qualification and the matches: it is the head starting to turn over the consequences. Anyone preparing a routine for the knockout stages is working on a problem of thoughts, not of arousal.

The second useful point is that the two knockout stages are equivalent. The authors expected the individual matches to be the heaviest stage, because the loser goes out, and instead anxiety before the team event was practically the same, 5.05 against 5.21. The idea that shooting as a three lightens the load, because responsibility is shared, does not show up in these data.

On the difference between men and women it pays to be precise, because it is easy to report badly. In this sample the women report more worry and more arousal than the men, and the gap is the same across all three stages: it is not that women react worse to rising pressure, it is that they start from a higher reported level and stay parallel. The study also measured neither trait anxiety nor a baseline away from competition, so it cannot say whether this is a difference in reaction to competing or a starting difference, and recalculating from its own tables the size of the effect comes out at about half the declared value. Above all: nobody looked at the scores. From these data it cannot be said that those reporting more anxiety shoot worse.

One last piece of context for reading your own experience: these are national-team archers, with eight years of practice on average, at a continental championship. Even at that level anxiety before a head-to-head match rises measurably, and it does so in both groups. It is not a sign of being unprepared: it is the stage that changes.