Psychobiology of archery: Landers 1986 study

Between 1982 and 1984, at four American venues including Olympic preparation camps and national championships, 188 recurve archers went through a two-and-a-half-hour battery of tests. There were 102 men and 86 women, aged from 12 to 74, with a mean age of 26.4 and a standard ranging from relative beginner to Olympic gold medallist. On each of them, 62 variables were collected across strength, body composition, vision, reaction times, attention, anxiety and mental habits, and related to the average of the official scores for the year in which the archer was tested.

Of the 62 variables, seven passed the threshold the authors had set themselves, a correlation above 0.25 at a probability below 0.005: thinking about past mistakes, relative leg strength, confidence, endomorphy, depth perception, reaction time and the use of visual imagery. Combined into a block model on the 79 archers with complete data, these seven explained 53% of the variance in score. The greatest weight lay in the first block, the mental one, at 26%: the better archers ruminated less over mistakes, used visual imagery with more command and reported more confidence. Then came relative leg strength and endomorphy at 14%, depth perception at 8% and reaction time at 5%.

The single correlations say the same thing more crudely. Leg strength relative to body weight was the physical variable most tied to the score, at 0.48; recovery from past mistakes and the use of visual imagery sat at 0.33, confidence at 0.32, endomorphy at minus 0.31. Grip strength gave nothing, with correlations of minus 0.03 and minus 0.01 for the two hands.

The block with the most variables is also the one that produced nothing: of the fifteen measures of attention and anxiety, none correlated with the score, and the highest stopped at 0.17. Trait anxiety sat at minus 0.05 and competition anxiety at minus 0.08. Cross hand-eye dominance did not separate the groups either, though there was a trend: none of the nine world-class archers was cross-dominant.

Dividing the 79 archers into four score bands, the differences run in the expected direction on almost everything: thinking about mistakes falls from 9.3 in the lowest band to 3.7 in the highest, relative leg strength rises from 1.24 to 1.72, reaction time falls from 194.8 to 173.5 milliseconds. The only variable that goes the other way is depth perception, which gets worse as the standard rises, and the authors offer two possible explanations: several of the top archers closed or squinted the non-aiming eye, and the better groups were on average six years older than the weakest one. Finally, classifying each archer above or below 1,200 points, five variables were enough to place them correctly in 81% of cases.

One last figure sits in a note: in a subgroup of 34 archers, the variability of reaction time from trial to trial correlated with the score at minus 0.33. The authors flag it as an indication to be checked, adding that consistency of response may count as much as its speed.

The archers were tested at Tempe in Arizona, at Colorado Springs, at Long Beach in California and at Oxford in Ohio, in single sessions of about two and a half hours, with the investigators trained beforehand to standardise the instructions. The performance criterion is not a score collected on the spot but the average of all the archer's official scores for the year in which they were tested.

The 62 variables fall into four families. The cognitive one comes from a questionnaire of 31 items on training and competition strategies, of which 19 were used, with free-text, multiple-choice and 0 to 10 scale answers. The physical one comprises nineteen measures: quadriceps and hamstring strength on an isokinetic machine at 60 degrees per second, three attempts per leg, with relative strength obtained by dividing the stronger leg's value by body weight; grip strength on a dynamometer, three attempts per hand; somatotype by the Heath-Carter method, which measures skinfolds, bone and muscle girths, height and weight to derive endomorphy, mesomorphy and ectomorphy; percentage body fat from three-site skinfolds; reaction time and anticipation time.

Reaction time was measured with a modified timer: the archer sat 1.8 metres from a trackway of light-emitting diodes, with a button in the dominant hand, and had to press when a red light came on between 1.0 and 2.5 seconds after a yellow warning light, the delay being chosen at random from four values. Twenty trials each, with the result given every time; trials below 120 or above 300 milliseconds were discarded and repeated. Immediately after each reaction the lights came on in sequence, simulating movement at 6.44 kilometres an hour, and the archer had to press again when they thought the last light would illuminate: that is anticipation time.

The visual part uses a standard screening apparatus for near and far acuity, phorias, depth perception and colour vision, plus Miles's cone test for eye dominance, crossed with the hand used for throwing and for drawing the string. The attention and anxiety part gathers five questionnaires already in use, among them the six attentional scales of Nideffer's test, a questionnaire on cognitive and somatic anxiety, one on the perception of autonomic signals, the trait anxiety scale of the State-Trait Anxiety Inventory and an attentional questionnaire developed for rifle shooting.

For the analysis, variables highly correlated with each other were reduced in two ways: by elimination, keeping only the one most correlated with the score, as among the seven leg strength measures, and by combination, summing conceptually close items, as with the two questions on past mistakes. The seven survivors were entered into a hierarchical regression in four blocks, with the order decided in advance by the authors on the logic of the research rather than by the statistics. Finally a stepwise discriminant analysis looked for the variables that place an archer above or below 1,200 points.

This is a cross-sectional, correlational study: everything was measured once and related to the year's score. None of the seven variables was manipulated, so nothing in this work says that training the legs or visual imagery raises the score. The authors write as much in closing: the findings need replicating on other groups before being used to select athletes or to build programmes. They also note that relative leg strength might signal not stability but the kind of person, that is, someone who keeps fit and perhaps also takes more care over equipment and practice.

The model was built on 79 of the 188 archers, those with complete data on all the chosen variables, and those left out are not described. The variables were selected by first looking at which ones correlated with the score in the same sample in which the model was then estimated: with 62 candidates and no check on new data, the share of variance explained is in all likelihood higher than it would be in another group of archers. The authors flag that problem in earlier studies, criticising their ratio of participants to variables, and in describing their own they write of "a ratio of subjects to predictor variables of 1:11", where the ratio is 11 to 1.

Several numbers do not agree from one part of the paper to another. The abstract speaks of 44 men and 35 women, the 79 of the model, tested between 1982 and 1984, while the methods describe 188 archers collected "over a two-year period". The four score bands add up to 81 archers in the results text and 80 in the table, where the model uses 79, and the hand-eye dominance analysis counts 81. Anticipation time appears in the table with a correlation of minus 0.26 marked as significant, that is, above the stated threshold, yet it does not enter the seven variables and the discussion describes it as uncorrelated. The endomorphy values reported in the tables run from 38 to 54, figures that do not sit on the Heath-Carter scale and that the paper does not explain; the discussion, moreover, describes the better archers as leaner, while directly measured percentage body fat did not correlate with the score, stopping at minus 0.06.

On reaction time, it should be kept in mind that trials below 120 and above 300 milliseconds were thrown away and repeated, so the distribution is cut at both ends, and that the measure is not taken on the bow but on a button pressed while seated, in response to a light. The difference of 22 milliseconds between better and poorer archers is the one between the highest and the lowest of the four bands; between the group above 1,200 points and the group below, the difference is about 7 milliseconds.

Then there is the paper's age. It is from 1986, the bows and arrows are those of the time, the reference score is the double FITA out of 1,440 and the psychological questionnaires are those of the 1970s. The score measure, finally, is a yearly average over different competitions, with different weather and different fields, and is not controlled for any of that.

The clearest thing this study says concerns the relative weight of the things measured. Among the seven variables left standing, the mental block, that is, ruminating less over mistakes, using visual imagery and feeling confident, weighed twice as much as the physical block, 26% against 14%, and depth perception and reaction time together contributed another 13%. It is a picture of associations, not a hierarchy of causes: in a group running from beginner to Olympic champion, those who shoot better also have a different relationship with their own mistakes.

The physical side yields two useful numbers pointing opposite ways. Leg strength relative to body weight was the physical variable most correlated with the score, at 0.48, and it rose in order from the weakest band to the strongest. Grip strength, by contrast, did not correlate at all; the authors explain this by the bow being held through two points of contact, the bow hand and the string anchored to the face, rather than one.

A set of negative findings is worth as much as the positive ones. None of the fifteen measures of attention and anxiety separated the archers, trait anxiety included: a finding that still bears on the habit of attributing scores to an anxious temperament. Cross hand-eye dominance did not separate the groups either, though with the curiosity that none of the nine world-class archers was cross-dominant. And depth perception went against expectation, getting worse as the standard rose, with two possible explanations offered by the authors: the non-aiming eye closed or squinted, and the higher mean age of the better groups.

On reaction time the finding is more fragile than it looks. Between the lowest band and the highest there are 22 milliseconds, but between those above 1,200 points and those below the difference drops to about 7, and the measure comes from a button pressed while seated, not from the clicker. More interesting is the note in the margin: in 34 archers, the consistency of reaction time from trial to trial correlated with the score almost as strongly as its speed. The authors flag it as a lead to be checked, and it is the same direction the later literature took.