The authors studied the mortality experience of nine economically advanced countries and found that over the past 100 years the modal age at death has increased substantially and variation among countries has diminished. In 1901, the modes fell within the range of 70 to 80 years. By 2001, the modes fell largely within the range 85 to 90 years.
The authors’ case for mortality compression, however, is based on a limited analysis of the distribution of deaths by age above the modal age. They found a decrease in the standard deviation of age at death among those who survived beyond the modal age from roughly 8.5 years in 1901, to roughly 7 years by 2001, on average. As with the modes themselves, the range of standard deviations of variation above the mode among the nine nations diminished from 1901 to 2001.
The authors’ calculation of standard deviation assumed a normal distribution using observations only above the mode, thus effectively implying that mortality compression has only to do with the distribution above the median. However, while the observation of compression of the distribution of age at death above the mode is valid, it appears to miss an important additional observation. The distribution of adult deaths by age below the mode was expanding as the distribution above the mode was compressing. This means that the symmetry of the distribution of adult deaths by age around the modal age has been changing, substantially. Rather than moving toward a symmetric, normal distribution as hypothesized by Fries, the distribution of adult deaths by age has become increasingly asymmetric, with the mode advancing more than the practical “maximum age” or omega. A better test of whether there has been compression of adult mortality might be to compute the standard deviation of age at death around the mode or the mean, using all observations of death above the age of 40. It appears that this standard deviation