Search

In the preceding issue of this journal, a generalized version of the Brass growth balance method was proposed that made it applicable to populations that are not stable and are open to migration. In this companion paper, the results of applying this new procedure to data from India's Sample Registration System for the decades 1971-80 and 1981-90 are discussed. The results at the national level show that, during the decade 1981-90, 5 per cent of the deaths among men, 12 per cent of the deaths among women, and about 7 per cent of births were being missed by the system. Further, it is estimated that the level of under-enumeration in the 1991 Census was more than that of the 1981 Census by 0.7 per cent for males and 1.4 per cent for females. The paper also presents results for major Indian states. [ABSTRACT FROM AUTHOR]

This paper presents a census-based method for estimating adult mortality. The method proposed here has several advantages over existing intercensal procedures. First, it is non-parametric in the sense that no model life-table system is required as in the other procedures; mortality conditions are inferred directly from the data. Nevertheless. when age-misreporting is severe, it may still be advisable to use the method in conjunction with model life tables in the manner that have been outlined. Secondly, the system is much easier to use than others, particularly when intercensal periods are not an integer multiple of five years. Thirdly, it is easy to use the estimation equations to identify the sensitivity of estimates to errors in the data, particularly to differences in completeness of coverage of the two censuses. The general advantages and disadvantages of intercensal estimation are well known. The main advantages are that data for using them are abundant and that no assumptions of stability are required.

Mortality decline has historically been largely a result of reductions in the level of mortality at all ages. A number of leading researchers on ageing, however, suggest that the next revolution of longevity increase will be the result of slowing down the rate of ageing. In this paper, we show mathematically how varying the pace of senescence influences life expectancy. We provide a formula that holds for any baseline hazard function. Our result is analogous to Keyfitz's ‘entropy’ relationship for changing the level of mortality. Interestingly, the influence of the shape of the baseline schedule on the effect of senescence changes is the complement of that found for level changes. We also provide a generalized formulation that mixes level and slope effects. We illustrate the applicability of these models using recent mortality decline in Japan and the problem of period to cohort translation. [ABSTRACT FROM PUBLISHER]

This paper reconstructs the trend in the population sex ratio in India between 1971 and 1996 from available information on changes in sex differentials in mortality in the country since the beginning of the century. It is estimated that, although the mortality of females relative to that of males in India has improved since 1968, the population sex ratio increased between 1971 and 1981, stayed constant between 1981 and 1991, and started to decrease only after 1991. This implies that the recorded decrease and increase in the periods 1971-81 and 1981-91 respectively were both spurious and were the results of undercounts of females in 1971 and 1991. Another implication of this finding is that, owing to the lagged effect of past mortality on current trends in the population sex ratio, this ratio is a bad proxy for use in the study of changes in differential mortality by sex. [ABSTRACT FROM AUTHOR]

Estimates of mortality in Cambodia during the Khmer Rouge regime (1975-79) range from 20,000 deaths according to former Khmer Rouge sources, to over three million victims according to Vietnamese government sources. This paper uses an unusual data source -- the 1992 electorate lists registered by the United Nations -- to estimate the population size after the Khmer Rouge regime and the extent of "excess" mortality in the 1970s. These data also provide the first breakdown of population by single year of age, which allows analysis of the age structure of "excess" mortality and inference of the relative importance of violence as a cause of death in that period. The estimates derived here are more comparable with the higher estimates made in the past. In addition, the analysis of likely causes of death that could have generated the age pattern of "excess" mortality clearly shows a larger contribution of direct or violent mortality than has been previously recognized. [ABSTRACT FROM AUTHOR]

Methods for correcting for underenumeration in mortality estimates have been developing intensively during the last fifteen years. While existing methods can be shown to be overspecified, this is particularly evident for techniques in which age-specific growth rates are used. The paper surveys some existing analytical techniques which do not use age-specific growth rates, by examining results and precision when used with 24 selected data sets. The specification problem is then analysed, and a new less-specified technique is introduced. The technique assumes an age-invariant rate of underenumeration of deaths, but allows age-specific growth rates to vary in a minimally consistent manner from a common general level. The results obtained by using this technique on the earlier data sets are presented, and the precision obtained is compared with the results from existing methods. The results are very favourable to the new technique. [ABSTRACT FROM AUTHOR]

Although Hungary is not alone in Eastern Europe in experiencing a rising death rate during recent years. this adverse development would seem to have progressed further there than in neighbouring socialist countries, with the possible exception of the Soviet Union. The Hungarian death rate has been rising since the mid-1960s in part because the population was ageing but, more significantly from the health point of view, because of a real increase in mortality among certain sections of the population. The age-specific death rates of males aged 15 and over were all higher in 1980 than in the mid-1960s, the increase being particularly marked for the age group 30-59: moreover, women aged 30-59 are also now beginning to display the same characteristic. In the paper the individual contributions of the various causes of death to these trends are examined and some of the factors that are thought to have enhanced the risk of dying are outlined. [ABSTRACT FROM AUTHOR]

Human populations have lived on the earth for millions of years, yet the study of population history only began to be established in the mid-twentieth century. In spite of the considerable progress in the study of historical demography which has since been made, there have been hardly any detailed studies of fertility and mortality before the sixteenth century. This study, by analysing a set of Chinese genealogies, examines long-term mortality patterns in a selected clan population over a period of more than 1000 years. The result shows that, in this selected population, mortality fluctuated around a relatively high level and showed no secular change over the very long period studied. The study also provides a comparison between the mortality patterns found in the selected population and those observed in a much larger Chinese lineage population, as well as those recorded among the British elites born between the sixteenth and the early nineteenth century. Based on the findings of this research, the paper presents some tentative suggestions about long-term mortality changes in Chinese history. [ABSTRACT FROM AUTHOR]

This article focuses on mortality in England and Wales. In principle, it is possible to make estimates of death rates at individual years of age in several different ways, depending on the information available about the numbers at risk. In some countries population registers are kept. In others, a census of population, supplemented by information about births, deaths, and migration since the last census, is used. However, in England and Wales it has been found that it is difficult to produce accurate estimates by this method at very high ages. During the early years of the present century, and indeed well before, there were considerable doubts about the accuracy of some of the high ages recorded in both the censuses and the death registrations. From 1951-55 onwards the death rates continued to fall. For males, they have fallen roughly in parallel in different age groups. For females, however, death rates have fallen rather faster at ages 80-89 than at ages 90-99. The reasons for this are not yet clear. It is possible that the death rates for females in 1951-55 may still have been understated at ages 90-99, more than they were for men and more than they were by 1980.

The article focuses on demographic factors and their fluctuation rates. Such rates are usually derived from a complete enumeration of the events rather than by sampling, and hence are not subject to sampling variations. Thus, it may be thought that observed rates provide a picture of the "true" situation. It is well known, however, that the smaller the population base on which such a rate is compiled, the more unstable is the rate over time. Statistically speaking these events are the outcome of a random experiment. These outcomes (such as birth, death, accident) are subject to chance. Thus, the observed rate may deviate from the "true" rate. Such deviation is called random error. If the experiments are repeated, a measure of this random error can be obtained by obtaining the average deviation of the observed rates around a "true value." In the development of statistics for small areas, it is needed to consider the smallest population which will provide useful information. In the conduct of field experiments, the investigator often selects small population aggregates as units of "treatment."

We use data from the 1931, 1941, and 1951 censuses of India and the 1951 census of Pakistan to examine the demographic consequences of Partition in the Punjab in 1947. Had growth rates for the period 1931-41 for the Punjab as a whole continued to 1951, the population of the Punjab would have been 2.9 million larger than that recorded in 1951. Population losses from migration and mortality above age 20 were approximately 2.7 million greater between 1941 and 1951 than would have been predicted by loss rates between 1931 and 1941. We estimate a net Partition-related population movement out of the combined Punjab of about 400,000. We conclude from several lines of analysis that Partition-related population losses in the Punjab, either from deaths or unrecorded migration, were in the range 2.3-3.2 million. Partition was also marked by a dramatic religious homogenization at the district level. [ABSTRACT FROM AUTHOR]

A persistent theme in much anthropological writing is the concept of the deliberate control of population numbers by hunter-gatherers as a means of achieving moderate family size, adequate nutrition, and constrained adult mortality. An analysis of the mix of theory and field evidence that led to this conclusion finds the case not proven. On the contrary, Malthusian constraints can operate, and probably did operate, to produce a hunter-gatherer society where most adults were reasonably robust and healthy even though child mortality was high and life expectancy short. The absence of population limitation in pre-Neolithic times implies high mortality as well as high fertility, and weakens the argument positing a Neolithic mortality crisis. [ABSTRACT FROM AUTHOR]

Data from the 1971 and 1981 census records of the OPCS Longitudinal Study (LS) members have been used to examine socio-demographic variations in the proportions of elderly people who lived in private households in 1971, but in institutions ten years later. Information on deaths of sample members 1971-85 has also been used as an indicator of the health status of various sub-groups of the LS population. The results show institutionalization rates to increase with age, to be highest for the single, and lowest for the currently married. Living arrangements in 1971 were also associated with differentials in institutionalization. Regression models which included a family/household variable fitted the data rather less well than models that included a marital-status term. Among men aged 65-74 in 1971, higher social class was associated with lower institutionalization. However, among older men aged 75 and over institutionalization rates were lowest for those from Social Class IIIM. An examination of 1971-81 mortality differentials lent no support to the hypothesis that this change in the relationship between social class and institutionalization might reflect selective survival effects. Differences in institutionalization by housing tenure showed that the lowest proportion in institutions in 1981 was found among owner-occupiers, while rates among private renters were the highest. Among men the institutionalization odds-ratio for private renters compared with owner-occupiers was 1.9 (95% confidence-interval 1.4-2.6), for women 1.3 (1.1-1.5). Between 1971 and 1981, mortality among local authority tenants was higher, and by inference health poorer, than among owner occupiers, so their higher rate of institutionalization might be expected. However, standardized mortality ratios for private renters in 1971-81 were lower than those for local authority tenants, implying better health, although rates of institutionalization were higher. An examination of variations in rates of institutional residence by availability of housing amenities suggested that housing-quality factors alone were unlikely to account for this. Differentials in the mortality between 1981 and 1985 of the 1981 institutional population suggested that mortality of those who had been private renters was lower than that of members of other tenure groups (although this difference was not statistically significant); these findings suggest that factors other than health status may contribute to the high institutionalization rates observed among private renters. Contrary to expectation, the mortality of the 1981 institutional population who ten years earlier had lived 'not in a family but with others' (generally relatives) was also low. Possible reasons for these findings are discussed. [ABSTRACT FROM AUTHOR]

Morbidity has increased during recent decades in Japan, the United States, and Britain, while mortality has declined. These patterns suggest that mortality decline may influence population composition in ways that contribute to higher sickness rates. In Hungary, mortality has increased since the mid-1960s. This experience, and results from the Hungarian health surveys of 1981 and 1986, make it possible to examine an inverse version of the hypothesis that sickness rates increase when death rates decline. Overall, and in most specific adult age groups, the prevalence of chronic disease in Hungary decreased. Hungarian experience in the 1980s suggests that population composition changed through the withdrawal of `non-survivors' individuals who lived shorter lives under the 1986 mortality regime than they would have lived under that of 1981. [ABSTRACT FROM AUTHOR]

This article focuses on the relationship between morbidity and mortality. Death is usually attributed to disease, and the attribution rests on the direct effect of disease on survival prospects. This aspect of the relationship leads to certain assumptions about how the incidence of ill health will change as the mortality rate changes. The most common assumptions are that high or low mortality rates are grounds for inferring high or low morbidity rates, and that the two measures change in a way that is parallel. That is, declining mortality rates are a sign of a transformation of health, which consists of declining morbidity rates. Except for virgin soil epidemics and the high infant mortality often observed in historical populations, most disease experiences end in recovery rather than in death. A rise in the average sickness of a population does not necessarily indicate a decline in medical well-being, because only the health of the surviving population is measured. If more people survive, then individuals who would have died live longer although their health remains poor. If we could compare individuals with equal endowments at birth, however, we would find an unambiguous improvement. The prospects of every individual may have improved over an earlier generation, even through average health has worsened.

This article focuses on researcher John Graunt and his contributions to the field of demography. Graunt, himself a merchant, was always busy with the bills of mortality, weekly and monthly tabulation of plague and other causes of death which merchants had long employed as a rough gauge of the propensity of their clientele to fly to the country in times of epidemics. Graunt had the idea that there should be inherent quantitative regularities in so many deaths, or "intrinsic and extrinsic values" as he called them. In a period in which even elementary terms of quantitative social reference such as "data" and "population" were not current, Graunt carefully evaluated the bills for their numerical consistency and reliability of compilation, and presented his evidence at length so that his readers might judge it independently. One needs to understand how Graunt arrived at these procedures, and what ideas and models of social collectivity he used to develop his arithmetic in the absence of both the term and concept of "population."

A pair of two-census methods of estimating mortality levels are tested with simulated census data. The populations considered range in size from 250 to 1500 individuals of each sex; censuses were taken at intervals of five and ten years. In general, the methods are resistant to bias, and yield variances similar in magnitude to those obtained using vital registration data and life table techniques for censored data. The two-census methods represent a substantial improvement over the techniques of mortality estimation previously available for small populations, since two reliable censuses are more likely to be available for these populations than complete vital registration. [ABSTRACT FROM AUTHOR]

This article presents information on the biological diversity of three populations living in the same area of Eastern Senegal, but with different histories and cultural patterns and to check whether some of the biological differences could be linked with cultural ones. The attempt to describe the demographic characteristics of each population and to compare these characteristics between populations forms part of this study. The collection of accurate demographic data was necessary for this purpose. The area chosen for this study is located in the Department of Kedougou in Eastern Senegal, near the border between Senegal, Mali and Guinea. It includes approximately half the total population of Bandafassi. The population of the area comprises three ethnic groups. A careful multi-round survey in small intensive studies can greatly reduce omission of events which occur during the period of observation. The risk of omission remains for children who die within a few days after birth, but is decreased by the registration of pregnancies at each round. Concerning underenumeration, subsequent surveys showed that the first census of the Fula Bande in 1975 missed at least seven per cent of the total population. Half the omissions were detected at the following rounds and the other half by the genealogical survey.

The relationship between smallpox epidemics, overall mortality and population growth, as reflected in an excess of births over deaths, has been examined with regard to the main sources of evidence from different parts of Europe. Epidemiological-demographic changes concomitant with different phases in the introduction of immunisation against the disease have been assessed in the light of evidence from records showing some causes of death as well as numbers of burials. By the eighteenth century; smallpox epidemics appear to have become predominant as an influence on fluctuations in overall mortality in much of Europe. Evidence is reviewed which suggests that although inoculation had probably protected many from smallpox after the mid-eighteenth century, to an extent that could have reduced overall mortality, vaccination enthusiastically promoted after 1800 had a dramatic epidemiological-demographic impact. Data from many sources have been summarised and indicate that the disease was virtually brought under control in North Western Europe during the course of the nineteenth century. Smallpox had probably caused between 8 and 20 per cent of all deaths directly in eighteenth-century Europe as well as unquantifiable secondary and associated morbidity and mortality. The removal of such a deleterious disease from a chain of infections affecting the population at this time, accounted for much of the increasingly more important role of mortality decline as the significant factor in demographic change. The evidence is circumstantial, but suggests that the unprecedented population growth of the early decades of the nineteenth century could in large part have been due to the control of smallpox through vaccination measures. The virtual elimination of the disease as a killer in Europe by the end of the century, following legislation and revaccination programmes was a unique achievement with further consequences for sustained population growth and improvements in health which for many were the only source of improvements in the standard of life. [ABSTRACT FROM AUTHOR]

A discussion of the surprising phenomenon of declining life expectancy in a highly developed country such as the Soviet Union during the 1970s shows that this result was probably due only in a small part to `true' causal changes in the conditions of living. At least equally important is the weaknesses of the measure of life expectancy by itself. The logical difference between period and cohort measurement is one part of the explanation. Another important factor is the adverse selection of risks by war which makes international and intertemporal comparisons less valuable. Factors like population redistribution or changes in the registration also contribute to the explanation. Thus; life expectancy (in particular period life expectancy) should not, without closer consideration, be accepted as a reliable indicator of human welfare under such circumstances. [ABSTRACT FROM AUTHOR]

A number of family reconstitution studies have been carried out in England in recent years under the aegis of the SSRC Cambridge Group for the History of Population and Social Structure. For this article the data from 13 such reconstitutions were pooled to enable the chief fertility, mortality and nuptiality characteristics of the population to be calculated (there is reason to suppose that the pooled data reflect national trends fairly closely). Results are presented for half-century time periods, 1600-49 to 1750-99. Marital fertility changed remarkably little over the two centuries but there were major changes in nuptiality and also substantial changes in mortality. [ABSTRACT FROM AUTHOR]

There is growing evidence of a substantial decline in infant mortality in England from the late seventeenth century onwards. This trend is examined in detail using data from the parish registers of a group of rural parishes in North Shropshire. A major change in the whole pattern of first year mortality during the period 1661-1810 is indicated, Its main features being an increase in mortality between the ages of six and eleven months, and a marked full in mortality during the first three months of life. Examination of the seasonal pattern of Infant mortality shows very heavy mortality among young infants in the winter, presumably from respiratory causes, during the period before 1700. It is suggested that a fall in the number of deaths horn these causes was the main reason for the decline in infant mortality since the late seventeenth century. [ABSTRACT FROM AUTHOR]

As a contribution to the debate on the influence of hospitals on mortality rates during the eighteenth and early nineteenth centuries this article suggests that the hospitals' role may have been underestimated. Part 1 begins with a consideration of the hospitals debate and the source materials available for a series of case studies of provincial hospitals and their patient catchment areas. Environmental conditions and demographic changes in seven provincial towns are examined focusing particularly upon mortality rates and the causes of death in order to gauge the scope for hospital treatment. Each hospital is then examined in turn, the initial conclusion being that the hospitals' physical layout and internal sanitary arrangements were usually planned with care during the late eighteenth and early nineteenth centuries. Only later, as population pressure on hospital resources grew, resulting in the overcrowding of wards and a rising proportion of serious or acute cases did conditions tend to deteriorate.

The article focuses on effect of mortality in a diversified group and the process to calculate it. In a homogeneous population the reduction and the extension are equal: a drop of one per cent in the death rate is equivalent to an increase of one per cent in the expectation of life. In a heterogeneous population, on the other hand, the reduction and the extension can be very different. Whenever statistics on the mortality of parts of a population are lacking, it is natural to assume homogeneity, in which case the expectation of life is calculated as the simple reciprocal of the probability of dying. In a homogeneous population, with death rate that is in which the chance of death for each individual per year is &b.mu; the chance that he will die in the first year will be &b.mu;. Suppose that, unknown to the observer, the population is not homogeneous, but consists of two groups, and that all those whose deaths are averted by an improvement of 100 δ per cent are not random members of the population, but belong to one group.

'Success', said Aunt Agatha in T.S. Eliot's "Family Reunion," 'is relative: it is what we can make of the mess we have made of things'. This aphorism has the ring of truth for anyone who has been concerned with the estimation of fertility and mortality rates in developing countries lacking reliable systems of vital registration. The messes which have repeatedly been made of things have led to the development of increasingly elaborate methods, both of data collection and of analysis. Substantial progress has undoubtedly been made, but he would be a rash person who would claim that a definitive solution has been found. The success is still relative. Undoubtedly the most complex and costly method of data collection is that known sometimes as the 'dual record system' and sometimes as 'PGE' after the Population Growth Estimation experiment of Pakistan.' Over a decade has now passed since the method was brought into prominence by the Pakistan experiment, and during that time several other surveys of this type have been conducted in other countries of the third world.

With the advent and accessibility of high-speed computers a little more than two decades ago, multivariate statistical analysis underwent a dramatic transformation. Suddenly, formerly prohibitive analyses requiring hundreds of hours of tedious calculations were possible in a matter of minutes, so that formerly arcane techniques of multivariate analysis became available to virtually anyone with access to a computer. The classic study of the relationship between economic development and fertility is that of D. Heer who, using multiple correlation, correlated male fertility with net national product per head and four other measures associated with economic development. These measures included infant mortality rate, newspaper circulation per head and two additional "control" variables - population density and percentage increase in energy consumption per head - chosen because previous studies had shown them to correlate with fertility. Heer apparently attempts to justify his use of partial correlation for causal inference by referring to certain "outside information" to establish the causal links in support of his hypotheses.

It is informed that most developing countries do not have the complete registration of vital events necessary for birth and death rates to be calculated directly with fair accuracy. However, knowledge of such measures has become necessary for the formulation of population and other policies. Mortality decline is the principal cause of the lack of stability in populations of many developing countries. This has led demographers to study the impact of mortality declines on the characteristics of populations, and to find a method to correct the estimates derived by the stable population technique from a population that is not stable because of mortality decline. In a population with declining mortality, the stable population technique overestimates the birth rate at younger ages and underestimates it at higher ages. Hence, the average of a number of stable estimates of the birth rate is nearly equal to the true rate because the negative and positive deviations cancel. It is opined that the stable population technique of adjustment does not require any supplementary information on the pace or period of mortality decline. At the same time, in the case of stable estimates of the birth rate, based on the ten-year intercensal rate of increase, which is commonly used, averaging also eliminates errors due to the age mis-statement.