[Digit preference in birth weight data of obstetric facilities].

An investigation on instrumentation of birth weight in obstetric facilities was made in the northern part of Fukuoka Prefecture from March 1989 to January 1991. A distribution of 3000 cases from 100 facilities was analyzed for the purpose of clarifying the reliability of the data. The following results were obtained. 1) Birth weights measured by analogue scales (1530 cases) showed digit preference in all distributions of the whole, the lower three digits and the lower two digits. But those measured by digital scales (1470 cases) did not show any distortion in these three distributions. 2) There were significant differences among mean birth weights calculated from the intact data, and 50g-, 100g- and 500g-interval frequency histograms in analogue scales (p < 0.01). But in digital scales there were no significant differences among mean birth weights. Rounding error was supposed to be one of the main causes of this discrepancy. Digital scales are superior to other types of scales because they do not show distribution distortion of the data. The reliability of the birth weight data is different among the types of scales. We have to consider which type of scale is used when we refer to annual changes or differences among areas in mean birth weights.