Analysis of repeated reported adult statures

A large sample of women reported their heights on two occasions with an 11-year time lag. The differences between the two reportings did not conform to the normal distribution hypothesis of measurement errors. The correlation coefficient between the two reportings was closed to the lower range of those between measured and reported heights. Mean reported stature was found to rise with time. The difference between the height reported at the two occasions could, when neglecting the miscoded cases, be described by a Laplace distribution.