General comparability of Registrar-General's reports

Edward Higgs

From 1837 in England and Wales, and from 1855 in Scotland, systems of civil registration existed that recorded births, marriages and deaths. These systems were administered by a General Register Office (GRO) in London and Edinburgh respectively, each headed by a Registrar General. The data on births, marriages and deaths published in the Annual and Decennial reports of the Registrar General for England and Wales, and for Scotland, are the most important source for understanding the long-term development of the demographic structures, and mortality experience of Britain in the nineteenth and early twentieth centuries. Historians and social scientists can use them to study demographic trends, epidemiology, the effects of the environment on mortality, and numerous other subjects. Since the Annual and Decennial reports give information by various administrative sub-units, as well as nationally, local and regional patterns can be studied over time and compared.

However, as with all historical sources, there are some problems of comparability over time. Those attempting to use information from the Annual and Decennial reports to identify trends need to be aware of these pitfalls, lest they confuse real changes with statistical artefacts. This is not to deprecate the attempt to study such trends, merely to urge caution and attention to secondary works of interpretation on the civil registration returns (e.g. Bryder; Glass, Hardy; Higgs, 2004a; Teitelbaum). Given care and an understanding of the limitations of the source, robust results can be produced. Many of the problems of comparability between the Annual and Decennial reports fall under five main headings:

1) changes to the questions asked at the time of civil registration;

2) changes to the known levels of under-reporting;

3) changes to the coverage of certain topics in the Annual and Decennial reports;

4) changes to the classification systems and other techniques used to analyse certain heads of information in the Annual and Decennial reports;

5) changes to the boundaries of the administrative units for which information is given in the Annual and Decennial reports.

Although the data collected at the time of registering a birth, marriage or death were fairly standard from the nineteenth century onwards, some changes had significant results. Thus, the use of the terms 'primary' and 'secondary' in relation to the cause of death information supplied by doctors on death certificates from the late Victorian period onwards allowed considerable latitude of interpretation. Doctors defined 'primary' in several ways, either chronologically, or in terms of the most important with regard to the termination of life. The subsequent introduction on the death certificate of the term 'contributory' as a variant of 'secondary' caused even more confusion, since it then became difficult to tell whether any 'secondary' cause was regarded as a consequence of the primary, or as of independent origin, but contributing appreciably to the death. These difficulties led the GRO to begin to doubt the very objectivity of the concept of a primary cause of death (Higgs, 2004a, 96). In 1927 the order of the statement on causes of death on the death certificate was changed by calling for the 'immediate' rather than the primary cause first, and then, in order, for any others of which it was the consequence. This then provided a more definite starting point for an analysis of the train of related causes (General Register Office, 1929, 145). The GRO believed that this innovation led to significant changes in reporting. There was, for example, an increase of 20 per cent in the number of maternal deaths registered as from non-puerperal causes associated with childbirth in the period after 1927 (General Register Office, 1938, 122).

Some care must also be taken when comparing data on births in the early decades of registration and those from the period after about 1880, because of evidence of early under-registration. This means that a comparison over time between published birth rates in the mid to late nineteenth century would give an erroneous picture of trends. D. V. Glass estimated that in the first decade of registration one has to multiply the number of registered births by approximately 1.080 to obtain the number of actual births. Teitelbaum used an inflation factor of 1.060 for the same period. By the end of the 1870s both agreed that birth registration was probably complete, especially after the passage of the 1874 Births and Deaths Registration Act (37 & 38 Vict., c.88), which introduced fines on parents for non-registration (Glass; Teitelbaum). Both marriage and death registration appear to have been have been virtually complete from the inception of registration.

In the Annual and Decennial reports the GRO did not always publish tabular data in exactly the same manner each year. Thus, in the nineteenth century the conventional age groups adopted in the Annual and Decennial reports were in quinquennia such as 0–4, 5–9, 10–14. Only sometimes were ages at death broken down into single years — 0, 1, 2, 3, and 4, or even more occasionally into months under one year of age (Woods, 37). This creates difficulties for the study of infant mortality prior to the inception of much more detailed reporting on infant mortality in the Edwardian period. Similarly, in England and Wales the regular tables showing the proportions of marks and signatures used by couples marrying in each registration district ceased after 1884; thereafter this information was only given for the larger registration counties and divisions. This creates problems for micro-studies of literacy levels.

Technical changes to the way in which data were analysed for publication in the Annual and Decennial reports can also cause problems of comparability. For example, as disease theories developed so causes of death might come to be placed under differing headings in tables. Thus, prior to 1869 deaths from typhus and typhoid were reported under the single heading 'typhus', only later being distinguished in the Reports. Similarly, diphtheria was originally categorized under the heading scarlet fever (Hardy, 478). Yet again, the development in the late nineteenth century of methods for recalculating mortality rates on the basis of standardised populations (i.e. controlling for changes or differences in age and sex structure) can cause problems for comparisons with the un-standardised figures in the earlier Annual and Decennial reports (Higgs, 2004b, 91).

Lastly, the administrative units in which data were reported in the Annual and Decennial reports also changed over time. In the period 1894 to 1898, for example, the General Register Office noted 797 changes to registration districts on the basis of which vital events were reported (Higgs, 2004b, 112). When establishing local trends it is important, therefore, to consult the footnotes in tables and local gazetteers to determine if administrative boundaries, and thus the population 'at risk', had changed. Further confusion was caused by the attempt in the period after 1911 to report data in terms of sanitary districts as well as in registration districts. The development in the late nineteenth century of mechanisms for returning deaths in large hospitals in London to the places where people normally lived may have increased the usefulness of the tables in the Reports, but may also have caused problems of comparability (Higgs, 2004b, 170–5).

REFERENCES

L. Bryder, 'Not always one and the same thing': the registration of tuberculosis deaths in Britain, 1900–1950'. Social History of Medicine, 9 (1996), 253–65.

General Register Office, Registrar General's Statistical Review for 1927 (London, 1929).

General Register Office, Registrar General's Statistical Review for 1936 (London, 1938).

D. V. Glass, 'A note on the under-registration of births in Britain in the nineteenth-century', Population Studies, 5 (1951–2), 70–88.

Anne Hardy, '"Death is the cure of all disease": using the GRO cause of death statistics for 1837–1920', Social History of Medicine, 7 (1994), 472–92.

Edward Higgs, 'The linguistic construction of social and medical categories in the work of the English General Register Office', in S. Szreter, A. Dharmalingam and H. Sholkamy, eds, The qualitative dimension of quantitative demography (Oxford, 2004a), 86–106.

Edward Higgs, Life, death and statistics: civil registration, censuses and the work of the General Register Office, 1837–1952 (Hatfield, 2004).

Michael S. Teitelbaum, 'Birth under-registration in the constituent counties of England and Wales: 1841–1910', Population Studies, 28 (1974), 329–43.

Robert Woods, The demography of Victorian England and Wales (Cambridge, 2000).