ARE STATISTICIANS LIARS?

[To the Editor of THE SPECTATOR] SIR,—Mr. Worsnop is mistaken in thinking that the figures he quotes about the average age at death of total abstainers and of moderate and heavy drinkers prove that the abstainer has no greater expectation of life than the moderate drinker. In fact, as I pointed out in my article, the figures prove nothing whatever about the relation between abstinence and expectation of life, for, at the time of the enquiry, total abstainers, as a group, were younger than non-abstainers, and, where two groups differ in age-composition, there is no point in comparing the average age of their members at death. The average age at death of people with false teeth is no doubt higher than that of people without them, but this does not mean that those with false teeth have a higher expectation of life ; it merely reflects the fact that, as a group, they are older than people whose teeth are their own.

The only accurate way of investigating the relation between abstinence and length of life is to compare the death-rate (or some other index of duration of life) among abstainers and non-abstainers of the same age. A large-scale investigation along these lines by Mr. R. M. Moore, an actuary, is described by Sir Arthur Newsholme (Vital Statistics, pp. 332-5). Moore's figures cover 290,226 abstainers and 338,262 non-abstainers, divided into thirteen five-year age-groups, from 20-25 to 80-85. If the death-rate of non-abstainers in each age-group is expressed as ioo, the corresponding figures for abstainers are:

Age-group. Cc mparative

death-rate.

Age-group. Comparative

death-rate.

20-25 70

55-60 71

25-30 70 60-65 78.5 30-35 56.5 65-70 84

35-40

54.5 70-75 86.5

40-45

57 75-80 ixo 45-50 58.5 8o--85 94 50-55

62

Moore's simple dichotomy of " abstainers " and " non- abstainers " may well obscure important differences between the death-rates of moderate and of heavy drinkers, and an enquiry with a finer classification would be of interest. But it is essential that it should be carried out on correct statistical lines.—Yours faithfully,

MARGARET KNIGHT.