Scientists generally know well the misuses to which citation metrics are often put. But they generally pay close attention to their own citation metrics, being keen to know that their own work, generated with such pain and angst, has been noticed.
Their attention embodies certain assumptions. Amongst these: that a paper which is cited has been read by the person who cites it, that it is cited for what it says, not for what it was assumed to say; and that it had an influence on the work of those who cite it.
In 1971, Carolus Oldfield published a paper entitled"The assessment and analysis of handedness: The Edinburgh inventory." (1) Being in Edinburgh myself, it seemed to me that perhaps I should be aware of this paper, which has been cited more than 20,000 times. Oldfield was a neuropsychologist who resigned a Professorship at Oxford to lead the M.R.C. Speech and Communication Research Unit in Edinburgh. I never met him, but I wish I had. The opening paragraph of his paper mentions “absurdities such as that of PARSON [I] who held that “ . . . handedness is caused by eyed-ness . . .“, and thus considered it necessary only to determine eye-dominance to establish the handedness of an individual.” Pity poor Parson, to see his name in capitals, but linked immortally with absurdity; we must blame the capitals on the Journal not the author, but it is nice to see such directness in scientific prose.
I looked at more of Oldfield’s papers. He was interested in thought and language, and he asked apparently simple questions that revealed themselves to have surprising subtlety.
For example, a letter to Nature, written with a PhD student, considered “The time it takes to name an object.” (2) They showed students pictures of objects whose names differed in how often they were used, and asked them to name the object drawn. Common names were retrieved faster than uncommon names. This is hardly unexpected; I don’t imagine that anyone would think the outcome of the experiment could be otherwise. But that seems beside the point, because just asking the question requires us to think about what how information is stored in the brain – how can the store of names be organised to allow common names to be retrieved most rapidly. Today we might frame an answer in terms of neuronal principles that were unknown in 1964, but it would still be a crude sketch of an answer.
But Oldfield and his student did the experiment and found something striking – that the speed of retrieval of the name shows a linear (inverse) association with the logarithm of the frequency of its occurrence in English texts (3). They did not know what to make of this, and said as much, but such simple regularities seem to speak of some simple underlying mechanism.
However, back to the Edinburgh Inventory. Oldfield had noted that handedness was not simple to assess, because which hand was dominant often differed according to the particular task. He therefore set out to devise a scale by which the degree of dominance might be quantified, with a questionnaire initially of 30 items that he simplified to a set of ten for convenience of use. As the figure below, reproduced from his paper, shows, very few individuals are completelyright-handed or left-handed. For him, this was a tool by which to quantitatively assess dominance in individuals with lesions to the right or left hemispheres of the brain, a tool that might enable the progress of recovery from such lesions to be assessed.
In the three years after its publication, the paper was cited a respectable (but unexceptional) twenty times. Interest picked up thereafter, and in the next ten years it was cited more than 400 times. But it has been cited more than a thousand times every year since 2009, and in 2018, 46 years after its publication, it was cited 1,509 times.
I did not read all of the citing papers. I looked at 15 randomly-selected papers published in 2018 from different groups - mostly brain imaging studies or studies of motor performance. Ten used the scale to exclude left-handed individuals, one to exclude right-handed individuals, and four merely reported the proportion of left-handed and right-handed individuals amongst their subjects, without subsequently using that information.
It thus appears that, while Oldfield, noting the limitations of any binary judgement of handedness, produced a continuous scale by which the degree of laterality could be assessed, many authors have used his scale to make a binary judgement of handedness.
Oldfield died in 1972, and what he might have thought of all this cannot be known. I suspect that he would have been pleased by the impact it had in the ten years after publication, when it generated an interesting literature, including serious critical reflections on the inherent problems of questionnaire data (4). He would probably have been surprised that it has become one of the most highly cited papers of all time. Perhaps he would have noted wryly that it seems to have been cited not because of what he had to say about handedness, but despite it.
Author: Gareth Leng | Date: February 2020
2. Oldfield RC, Wingfield A.(1964) The time it takes to name an object. Nature. 202:1031-2.
3. As given in the Thorndike-Lorge Word Count – a list compiled in 1944 of 30,000 words in English and their frequencies of occurrence. This extraordinary project, undertaken before computers, involved counting 18 million instances of the use of 30,000 words in English texts. The project gave gainful employment to many students in the 1930s.
4. Raczkowski D et al. (1974) Reliability and validity of some handedness questionnaire items. Neuropsychologia 12:43-7. “Questionnaires of this type are not without their problems. Principally, it is unsafe to assume that a subject who says he uses one or the other hand for a particular task actually does so. His memory may be inaccurate, or he may show a “halo effect,” answering uncertain items the same way he answers most other items.”