Emotions and scores in archers: Wang 2022 study

Sixty-one athletes from the Chinese national archery team, 26 men and 35 women, completed three questionnaires and a battery of attention tests; a week later they competed, and the score from that competition is the performance measure. The technical levels, assigned by the Chinese sports administration, were 12 at international level, 30 national and 19 first-class; mean age is 21.39 ± 3.51 years and training years 8.93 ± 2.94.

The three psychological variables are sport-confidence, measured both as a state and as a trait, attention, measured with three paper-and-pencil tests of three minutes each covering attentional allocation, breadth and shifting, and cognitive reappraisal, that is the habit of reframing a situation to change how it feels. The model was estimated with partial least squares structural equation modelling.

The model explains 33.6% of the variance in competition score. Sport-confidence is the strongest predictor, with a total effect of 0.574 (p < 0.001, confidence interval from 0.342 to 0.796), followed by attention at 0.344 (p = 0.009, interval from 0.068 to 0.581). Attention partially mediates the relationship between confidence and performance (0.130).

On cognitive reappraisal the result is two-sided. Its direct effect on the score is negative, −0.268 (p = 0.020), but passing through sport-confidence the indirect effect is positive, 0.202, and the two cancel: the total effect is −0.007 with p = 0.964. The authors call this a suppression effect and draw a recommendation from it: cognitive reappraisal helps regulate emotions and raises confidence, but using it during competition might consume cognitive resources, so it is better to work directly on confidence. On the path towards attention, reappraisal acts only through confidence (0.172); the direct link with attention was not significant and was removed from the model.

The design is cross-sectional, with the questionnaires administered to the group all at the same time, two hours after the end of training, in a quiet, well-ventilated and warm room. The competition took place a week after the tests, under Olympic rules. Informed consent was collected from athletes and coaches, and the technical level data were taken from the athlete qualification management system before the competition.

Performance is an index built by the authors: the competition score, meaning the total from the qualification phase plus the total of sets in the elimination phase, together with a score for the athlete's official technical grade.

Attention is measured with the Elite Athlete Attention Test, a paper-and-pencil battery with three subtests of three minutes each. In the first, two target shapes have to be found and marked among sixteen similar shapes, on a sheet of 300 figures, and it measures attentional allocation. In the second, the squares containing four small circles have to be marked among squares with different numbers of circles, across 650 squares, and it measures breadth. In the third, additions and subtractions alternate with the result written between two numbers, across 264 results, and it measures attentional shifting. The examiner explained each subtest and checked that it was understood before starting.

Sport-confidence is measured with two instruments: the state sport-confidence inventory in its Chinese version, 13 items on a nine-point scale, and the athlete sport-confidence inventory, 16 items on six points, measuring trait confidence in its task and coping dimensions. In this sample internal consistency is 0.94 for both. Cognitive reappraisal is measured with the subscale of the same name from the emotion regulation questionnaire in its Chinese revision, six items on seven points, internal consistency 0.80 in this sample.

The analysis uses partial least squares structural equation modelling in SmartPLS 3.3.3, chosen by the authors because it tolerates small samples, does not require normally distributed data and is oriented towards building a theory rather than verifying one. The significance of coefficients comes from bootstrapping with 5,000 resamples; predictive relevance from blindfolding with an omission distance of 7. To respect the rule on minimum sample size, the state and trait confidence scores were each summed into a single indicator.

The final model is not the starting one. Two of the six cognitive reappraisal items were eliminated because their loadings were 0.475 and 0.586, below the 0.6 threshold, and the hypothesis of a direct link between cognitive reappraisal and attention was removed because the coefficient was not significant. The results reported come from the model as thus revised.

The model was revised by looking at the data, and then estimated on the same data. Two of the six cognitive reappraisal items were eliminated for low loadings, and one of the hypothesised paths was removed because it was not significant. This is common practice in this kind of analysis, and the authors declare it, but it has a consequence to keep in mind: the fit, coefficients and p-values of the final model are more favourable than they would be on new data, and the 33.6% of variance explained is not an independent estimate. The most heavily adjusted variable is precisely the one on which the main result rests.

The sample is 61 people for a model with four constructs and several mediation paths. The confidence intervals show it: the total effect of attention runs from 0.068 to 0.581, that is from almost nothing to large, and that of confidence from 0.342 to 0.796. These are estimates compatible with many different scenarios.

The result on cognitive reappraisal, the one that gives the paper its title, is the most fragile of the three. It rests on a suppression configuration, that is a negative direct effect (−0.268) cancelled by a positive indirect effect (0.202) through confidence, with a total effect of practically zero. Configurations of this kind are unstable in small samples and depend on which paths sit in the model, and here one path was removed after being tried. The scale has also lost a third of its items. The explanation the authors give, that reappraising consumes cognitive resources during the shot, was not measured in any way: there is no index of cognitive load, and the questionnaire records habitual use of reappraisal in daily life, not what the archer did during that competition.

Technical level does not enter the model, although it was collected, and in this sample it is closely tied to age and years of practice: the international-level women are 27.17 ± 2.86 years old with 13.33 years of training, the first-class men 16.50 ± 1.29 years old with 5.75. The sample therefore runs from adolescents to adults, and confidence, attention and score all rise with level: part of the link between confidence and performance may reflect how far along each athlete is rather than an effect of confidence. The same holds for the two sexes, analysed together.

The performance measure combines two different things: the qualification total, which depends on the archer alone, and the total of sets in the matches, which also depends on who you meet and how many matches you shoot, since those knocked out early have fewer. Of the model's four variables, performance is the one with the lowest reliability, with a Cronbach's alpha of 0.696, just under the 0.7 threshold the authors themselves cite as a criterion.

The design remains cross-sectional, and the authors put this first among their limitations: the model's relationships cannot be read as causal. One point must be granted this study that many similar works lack, though: the questionnaires were completed a week before the competition and performance is a real competition score, not a self-rating nor a laboratory test. Among the other limitations the authors declare: they could not collect data at several time points nor run an experimental study, the psychometric properties of the measures need further verification, psychological measures alone are not enough without technical and fitness constructs, and the model applies for now only to an elite population. The model fit, finally, is described as good while the value of 0.336 falls, on the scale the paper itself cites, in the medium category.

This is one of the few psychological studies in archery in which performance is a real competition score and the psychological measures were taken beforehand, a week ahead. That is why its orders of magnitude are worth something. Among three mental variables, sport-confidence has the strongest link with the score (total effect 0.574), followed by attention (0.344). Together they explain a third of the variation in scores, which is a good deal for questionnaire measures, and leaves two thirds elsewhere.

A useful detail about how confidence is measured: the authors record it in two ways, as a state and as a trait, and keep the two together. In their conclusion, work on confidence has to address both. The same holds for attention, measured across three distinct aspects, allocation, breadth and shifting, with three different tests: if attention is to be trained, this study suggests it is three things rather than one.

On cognitive reappraisal, that is the habit of reframing a situation to change how it feels, it pays to stop at what is there. In the data, those who use it more tend to have more confidence, and that goes with better scores; but the direct link with the score, at equal confidence, is negative. The overall effect is zero. It is an interesting result and fragile enough not to make a rule of it: the scale lost two of its six items, one path in the model was removed after being tried, and the sample is 61 people. The idea the authors propose to explain it, that reappraising while shooting takes up resources the shot would need, was not measured: the questionnaire asks how emotions are managed in life, not what was done between one arrow and the next.

One caution about the sample: within these 61 athletes there are international-level archers with thirteen years of training and sixteen-year-olds with fewer than six, and level was not held aside in the analysis. Since confidence, attention and score all rise with years of practice, part of what the model attributes to confidence could simply be experience.