In 2018, 100 years after the ending of the First World War, I read the Haig’s War Diaries. The editors had provided in footnotes a potted history, with the years of birth and death, of those mentioned in the text. Most were senior officers, brigadier-generals and above, but some were colonels and a few were majors. What struck me as I turned each page was that many had lived to a ripe old age. But did they live, on average, longer than the man in the street? I decided to find out since I did not know what to expect. Social class is known to be associated with greater longevity and these officers, brought up in wealthy families during the 19th century, were, for want of a better term, posh. However, being a senior officer in the First World War and in previous campaigns they would have served in, was not exactly a stress-free existence. Over 200 generals were killed, wounded and taken prisoner; many were sacked if they were thought not to be performing well.
For each British or Empire officer mentioned in the footnotes, I calculated the age at death (taken here as the year of death minus the year of birth, see below). I excluded any killed during that war. The average age at death, for the 247 senior officers (referred to from here as ‘Generals’) was 75 (median 76). But what about the general population?
The calculated lifespan was not that from birth; it was from men who had survived until 1915 and who had been born between 1850 and 1886 (median 1867). I could find no life tables used by insurance companies, for example, for the general male population born around 1867 and who had reached the age of around 50. Life tables for people born at different times cannot be used because average longevity has changed over the decades. So I decided to make my own. The procedure I used to obtain a random sample, using the database for civil deaths in England and Wales on findmypast.com is described in an annex below. I used a range of dates of birth, around the median for the Generals of 1867, equal to plus and minus one standard deviation (8), i.e. 1859 to 1875.
For English and Welsh men born between 1859 and 1875 and who lived until at least 1915, the average and median age at death was 72. The average General outlived the man in the street by about 4 years.
For those interested in statistical calculations, the probability for differences in mean and median, P, was <0.0001. In other words, there is less than a 0.1% probability that the difference is due to chance.
There is a small difference in the way the age at death was calculated. For the Generals it was obtained by subtracting the year of birth from the year of death which leads to an overestimate of, on average, 6 months. Data for the whole population were from the age at death reported to the registrar. However, having looked at hundreds of death certificates, there are many cases where the informant did not know the exact date of birth and the age recorded is that I used for the Generals. Whatever the true error, in round terms the Generals lived 4 years longer.
The graph below shows the survival curves for the Generals compared with the general male population. It is evident that the difference in longevity of around 4 years was established at a relatively early stage. The death rate of the general population was higher from the age range of 50 to the early 60s. The extent of the difference can also be seen in terms of survival to a certain age. For example, by 75 50% of the Generals were still surviving compared with 38% of men in the general population.
Douglas Haig himself, included in the data set, was not one even reaching the average age of death; he died at the age of 66.
But if First World War generals lived on average of 4 years longer than the man in the street, what about the longevity of men in other professions? That question I look at in the next article.
I have had helpful discussions with Professor Tom Kirkwood and Chris Daykin CB, formerly Government Actuary.
Haig D. 2005. War Diaries and Letters 1914-18 (edited by Gary Sheffield and John Bourne). Weidenfeld & Nicolson.
---------------------------------------------
Annex: Constructing the Life Tables for the Male Population of England and Wales
The life tables were made for men born between 1859 and 1875 who died aged over the age of 40.
On findmypast.com the index of death records (Civil Deaths and Burials) appears on the screen on pages of 20 names. In order to obtain as near to a random sample as I could devise, I used the following procedure to extract data.
I constructed one set of tables as follows:
I searched for all surnames beginning with ‘A’ (i.e. A**** as the search term). On the first page that came up, I took the age at death of the first male looking upwards from the bottom of the page who fitted the criteria. If no male in the date of birth range was on that page, I went to the next page until one was found. I then jumped to page 10 and repeated the process. When the ‘A’ surnames were exhausted, I moved on through the alphabet until I had collected the same number of records as I had for the Generals (247).
Being slightly concerned that there could still be a bias towards names near the beginning of the alphabet, I then repeated the whole process, this time beginning at the end of the alphabet and jumping by 1 screen page at a time rather than 10.
For the first group the mean age at death was 71; for the second 72. The standard deviations were such that there was no statistical difference between the two groups. The two were therefore combined into one ‘Male Population’ sample
Then as a final check, I extracted the ages at death from a single registration district of all the men (269 in total) born in 1865 who died in or after 1915. The mean age at death was 71.3.
In a life table for men born in 1870 sent to me by Chris Daykin, the median age at for those surviving beuond the age of 50 was 72., supprting my view that I had achieved a random sample from the procedures I used to extract data from findmypast.com.
My tables, deaths from the age of 40, explain why the survival graphs from the age of 50 do not start at 1.00 (i.e. 100%).
No comments:
Post a Comment