EMG and archer skill level: Ertan 2005 study

Twenty-three people in three groups. Seven archers of the Turkish national team, with a FITA score of 1303.4 ± 26.2 out of a maximum of 1440. Six beginners from a city club, at 1152 ± 9.0. And ten university students who had never shot: for these ten the score was not measured at all, the authors assigned it as 250 for all ten, the same figure with no variability, after asking four national team coaches how much someone with no experience might score and after seeing that a beginners' class, following eight weeks of lessons, averaged 288 with a standard deviation of 65.

Electrical activity was recorded for each person from the finger flexors and the finger extensors of the forearm on the drawing side, over the two seconds around the fall of the clicker. From those traces the authors derive four indexes, each the mean area under the processed curve within a chosen time window: from 220 milliseconds before the clicker to 80 after, on the extensors, is the first; from 160 to 440 after, again on the extensors, is the second; the two intervals added together, 580 milliseconds in all, is the third; from 240 to 440 after, on the flexors, is the fourth.

Since the scores were not normally distributed, the authors turned them into ranks from 1 to 23 and regressed the rank on the natural logarithm of each index. The four correlations are 0.66 (p<0.0008), 0.70 (p<0.0003), 0.74 (p<0.0001) and 0.63 (p<0.002). The text and the abstract give them without a sign, while the figures and the discussion make them negative: the fitted lines slope downwards and the discussion states plainly that more experience goes with less area under the trace.

The authors read this as the trained archer not holding the weight of the string in the fingers but spreading it across forearm, arm, shoulder girdle and back, so that the share taken by the forearm falls. The peak that appears about 260 milliseconds after the clicker also drops as experience rises. The conclusions propose the indexes for assessing technique, for following progress and for selecting talent.

One session per participant. Three practice shots to get used to the equipment, then twelve shots in a row recorded, and for the calculation the mean of the 8 to 12 shots left after the noisy ones were discarded.

Surface electrodes of silver and silver chloride, skin shaved, lightly abraded and cleaned with alcohol, two centimetres between the centres of each pair, reference electrode on the olecranon of the drawing arm. Signal pre-amplified at a gain of 500, band-pass filtered between 8 and 500 hertz, sampled at 1000 hertz and converted with a 12 bit converter. A mechanical switch under the clicker marked the instant it fell, and that marker was superimposed on the electromyographic trace on the same time scale. Each recording lasted five seconds, two and a half before and two and a half after the clicker, and the window used for the calculations is the second before plus the second after.

The signal was rectified, averaged and smoothed with a moving-average filter over a 40 millisecond window in the published version, then normalised to each person's maximum voluntary contraction, obtained by pulling against a fixed resistance with the three-finger hook, at the same finger joint angles as in shooting. The indexes are therefore percentages of that person's maximum contraction, not absolute values.

The boundaries of the four time windows were not fixed in advance: the authors chose them by maximising the squared correlation coefficient between the rank of the FITA score and the index. The scores of the two archer groups came from official competitions, over 144 arrows at four distances, 90, 70, 50 and 30 metres for men and 70, 60, 50 and 30 for women. The ten without experience shot no round at all, so as not to expose them to the risk, and their 250 comes from the interview with the coaches. The Kolmogorov and Smirnov test showed the scores were not normally distributed, hence the transformation into ranks with ties replaced by the mean rank; the relationship looked exponential, hence the logarithm of the indexes.

The study was funded by a research grant from the Middle East Technical University. Participant age, the sex composition of the groups, the draw weight, the bow type, an ethics approval statement and a limitations section are all absent.

The limit that weighs more than any other sits in one line of the methods: the four time windows were chosen by maximising the correlation that is then reported as the result. There is no validation on data other than the data used to pick the boundaries, no group held back, and no correction for the fact that the four indexes come from the same two signals on the same twenty-three people. The 0.74 is therefore the most that can be obtained by fitting the windows to this sample, not an estimate of how the index would behave on a new archer, and the p-values as written cannot be read as probabilities.

The second limit concerns the variable being explained. Ten of the twenty-three scores are not measurements: the ten without experience did not shoot, they all carry the same 250 and a standard deviation of zero. After the rank transformation those ten share a single tied rank, so the twenty-three points of the regression rest on fourteen distinct values and, in practice, on three levels. The correlation largely measures the distance between people who have shot for years and people who have never shot, which is a distance known before the electromyograph is switched on.

The third point follows from that. The difference a coach would care about is the one between the seven national team archers and the six beginners, and it is not reported as a comparison in its own right: two groups of seven and six people, with scores of 1303 ± 26 against 1152 ± 9 that do not overlap. Without that comparison the conclusion about talent selection does not hold: an index that separates archers from non-archers tells a coach what they find out by asking how long someone has been shooting, and nothing here shows that it ranks archers of the same level against one another, which is what selection requires. In the same way, the proposal to use the indexes to follow progress is never put to the test: there is one session only and nobody was followed over time.

Score and electromyography were not collected together. The scores come from official competitions on unreported dates, the muscle activity from a single laboratory session: how much time separates the two is not stated, and in the meantime an archer's form moves.

On the measurement itself, the authors warn that the two muscles sit among others and that the activity was picked up as forearm flexors and extensors taken as a whole: the index is a coarse forearm signal, not a signal from those two muscles. Normalisation runs through a maximum contraction using the three-finger hook against a fixed resistance, which is itself hard to standardise, and an error there shifts the whole index.

The FITA score is also put into a single ranking across men and women, who shoot different distances: the same number does not mean the same thing, and the sex composition of the groups is not reported. On the sign, anyone stopping at the abstract or at the results section would take the relationship the other way round, because there the four correlations are written without the minus sign that the figures and the discussion show. One difference between the two versions of the text in circulation remains: the uncorrected proof gives a smoothing window of 80 milliseconds, the published version 40, and anyone wanting to rebuild the indexes needs to allow for that.

What an archer takes home from this study is a direction, not a number. In the forearm of the drawing hand, less electrical activity goes with a higher score, and in the seven national team archers the peak arriving just after the clicker is lower as well. The authors read this as the weight of the string, in the trained archer, staying out of the fingers and spreading across forearm, arm, shoulder girdle and back. Anyone who has shot for years recognises the description: the hook that does not grip and the release that comes from letting tension go rather than opening the fingers.

What the study does not give is a value to aim at. The four indexes are areas under a curve normalised to each person's maximum contraction, within time windows cut to fit these twenty-three people: there is no reference value, there is nothing on how much the index varies from one day to the next in the same person, and it is not shown to fall when technique improves. The selection idea finds no support here either, because the difference reported is the one between people who shoot and people who never have.

For anyone using electromyography as a working tool, the value of the study lies elsewhere: in the details of the protocol, which are written out precisely, and in the idea of anchoring the analysis window to the fall of the clicker with a mechanical switch, which gives a sharp time reference in a movement where the moment of release is hard to pin down. The registry keeps this work as a cross-sectional comparison of three levels in which one level's score was assigned rather than measured, and in which the windows of the index were chosen to maximise the association then reported.