Abstract
The BBC television programme Right Hand, Left Hand, broadcast in August 1953, used a postal questionnaire to ask viewers about their handedness. Respondents were born between 1864 and 1948, and in principle therefore the study provides information on rates of left-handedness in those born in the nineteenth century, a group for which few data are otherwise available. A total of 6,549 responses were received, with an overall rate of left-handedness of 15.2%, which is substantially above that expected for a cohort born in the nineteenth and early twentieth centuries. Left-handers are likely to respond preferentially to surveys about handedness, and the extent of over-response can be estimated in modern control data obtained from a handedness website, from the 1953 BBC data, and from Crichton-Browne's 1907 survey, in which there was also a response bias. Response bias appears to have been growing, being relatively greater in the most modern studies. In the 1953 data there is also evidence that left-handers were more common among later rather than early responders, suggesting that left-handers may have been specifically recruited into the study, perhaps by other left-handers who had responded earlier. In the present study the estimated rate of bias was used to correct the nineteenth-century BBC data, which was then combined with other available data as a mixture of two constrained Weibull functions, to obtain an overall estimate of handedness rates in the nineteenth century. The best estimates are that left-handedness was at its nadir of about 3% for those born between about 1880 and 1900. Extrapolating backwards, the rate of left-handedness in the eighteenth century was probably about 10%, with the decline beginning in about 1780, and reaching around 7% in about 1830, although inevitably there are many uncertainties in those estimates. What does seem indisputable is that rates of left-handedness fell during most of the nineteenth century, only subsequently to rise in the twentieth century.
Acknowledgements
We are extremely grateful to Dr Chuck Wysocki for providing the raw data from the Gilbert and Wysocki study.
Notes
1It should be noted that the loess curve is only a first approximation to the best-fitting line because both (a) no weighting is applied according to the size of the various data sets, and (b) it is a theory-free description of the data, unlike the Weibull function.
2The regression equation described below would predict that there should be about 4.65× as many left-handers found compared with the general population, whereas there were actually 10.44×, about 2.2× as many as expected above the normal excess of left-handers. We suspect that some parents either encouraged their left-handed children to respond or perhaps even responded on their behalf.