ArticlePDF Available

Abstract and Figures

Religion and its effects on morbidity and mortality (with particular emphasis on mortality) are reviewed as are special issues which have in the past made the study of religion and death difficult. The morbidity and mortality experience of various religious groups is portrayed, including Protestants, Catholics, Jews, Muslims, the clergy, Seventh-day Adventists, Latter-day Saints, Parsis, Jehovah's Witnesses and Hutterites. Studies of religious effects on morbidity and mortality have broadened in focus from the study of specific health practices, or health-related behaviors, to include the study of social support, religious participation and health-related attitudes. Gaps in the literature are identified and a preliminary model of religion's effect on morbidity and mortality is discussed.
Content may be subject to copyright.
Religiosity and Health
K.Srinivasan1 and Raka Sharan2
Indian Institute of Technology, Kanpur
There are studies showing linkages between religious beliefs and health. The
importance of religion on health has been examined by studies (Vaux 1976), the
dietary beliefs in health and illness (Chan Ho 1985), the role of religion in morbidity
and mortalitiy (Jarvis and North Cott 1987), the Religion and other factors influencing
health status (Idler and Kasl 1992).
In India studies showing the role of caste and religion in terms of specific behaviour
of individual groups in the power structure and related this to their healthculture
(Banerji 1982), and the role of magic and other beliefs among a south Indian
caste in health (Dumont 1986). Based on the concurrency in the issue the present
paper focuses its attention on the various aspects of religiosity and health
It is observed by various researches that the rural population of India, is very
much influenced by religious beliefs. For example cultural formation of individuals
closely inter linked with performance of individual's daily routine. Therefore, iet us
first examine the impact of religious practices and rituals as aspects of religiosity on
health with specific reference to rural individuals of India
Research Methodology
The rural population of Tamil Nadu is selected as universe of the present study
because of its spectacular achievements in the health administration. Since this
study is of health-culture of rural population, two villages were chosen for making a
comparative research design. The total population of both the villages were
interviewed for collection of desired information.
Data Collection
For data collection, a set of interview schedule was used. It was concerning with
the health factors which was administered on the heads of the households who are
the major respondents of the study. Some of them were close ended and some
were open ended. Totally 207 the heads of the house holds were interviewed. The
total period of data collection was spread into 7 months i.e. between November
1992 and January 1993, and between April 1993 to July 1994. The spell of field
study was not very smooth. For example, many respondents were initially reluctant
to answer some of the questions. However, such cases were overcome by
persuasion through personal contacts and rapport.
1 Presently working in Indian Institute of Information Technology and Management Kerala, Technopark,
Trivandrum 695581 Email
2 Professor(Retired).
Considerable care has been taken to ensure that the data collected for this study were as
objective as possible, and they have been collected they formed the basis for interpretation
of phenomena observed and for drawing conclusions. There are three types of scales
developed. One is Socio Economic Status scale( SES ). The second type of scale which is
used is on Ritualism.
And the third type of scale is on Health. The Socio Economic Scale was developed with
caste, income, occupation, education and age. A set of judges who are familiar with the
area were asked to rank the caste and occupation, based on the rank given, and
respondents are categorized into three categories i.e. Low, Middle, and High.
Age was made into three-point scale. In case of age, the respondents below the age
group of 30 are categorized as Lower(l), the respondents between the age group of 30
and 45 are middle age(2) and 45 and above are higher age(3).
Income scale was mad based on the following criteria. Income ranges from Re.O to Rs.300
per month placed in Low(l), from Rs. 601 to Rs.2000 placed in middle(2), and 2000 and
above placed in higher income(3).
Education the respondents who are in the category of no education and Primary level of
education(i.e. Five years of schooling), are categorized as Less educated; the respondents
in the secondary level (i.e. above five years of schooling and up to twelve years of
schooling) are categorized as middle level of education and the respondents who are
educated the University level of education and above are higher level of education.
Religiosity was measured by performance of various kinds of ritualistic actions. On the basis
of the observations during the field work, the following items as indicators to measure
the levels of religiosity and levels of health were developed.
To determine the degree or extent of the respondent's religiosity the following four
indicators were selected:
(i) Visit to place of worship
(ii) Celebrating religious ceremonies
(iii) Performance of rituals
(iv) Restriction on dietary practices.
Respondents who visited to the places of worship daily were assigned 3 points, the
respondents who visited to places of worship once in a week were assigned 2 points, and
the respondents who visited to places of worship once in a month or occasionally were
assigned 1 point. For the questions on religious ceremonies and rituals, 1 point to each
positive response and zero value for negative responses were given. For the respondents
who kept fast and maintained restriction on diet at least once in a week were assigned 3
points, the respondents who kept fast once in a month and some restrictions on diet
were assigned 2 points, and who kept fast once in a year or occasionally and who
maintained occasional restriction on diet on some specific days were assigned 1 point.
The total score ranged from 2 to 8. Those who secured 2 points were placed in
'Less religious' category, those who scored 3 to 5 points were place in ' Moderately
religious' and those who secured 6 to 8 points were placed in 'highly religious' category.
Thus, the scale was divided into three major categories: less religious, moderately
religious, and highly religious.
Likewise, to assess the extent of 'good health' the following five indicators were used:
(i) Incidence of sickness,
(ii) use of physical health measures,
(iii) paying attention towards personal hygiene,
(iv) consumption pattern, and
(v) sanitation
For all questions 1 point to each positive response and zero point to each negative
response were assigned. The aspect of sickness was inclusive of frequency, type and
The total score ranged from 0 to 5. Those who secured up to 1 were placed in
'less healthy' category, those who scored between 2 and 3 points were placed
in 'moderately healthy' category, and those who secured 4 and 5 points were placed in
' highly healthy' category. Thus, the scale of health-status was divided into three major
categories: less healthy, moderately healthy and highly healthy.
Some studies have already pointed out some relationship between moral conduct of
individuals and health (Cartstairs, 1965, Hasan, 1967). The studies reported that the roots
of illness extend into realm of human conduct and cosmic purposes. Further, these
studies have mentioned that villagers did not pay attention on their health care but they do
care to follow certain practices in a very rigid manner. For example, the villagers are in
habit of taking early morning walk either for a dip in the holy rivers or toilet purposes,
following of certain kind of restrictive diet on certain days; keeping fast on certain specified
days etc. All of these hygienic and health practices are linked with the aspects of religiosity.
Likewise, the habit of bare-footed trekking and of smoking from the same hobble-bubble
are some of the in unhygienic traditional practices directly affecting the health (Carstairs,
1965, Hasan, 1967) These habits are known as religious practice; and they have roots in
the frame-work of religion.
In sociological tradition, religion is considered as an institutionalized system of
symbols, belief values, and practices. Thus, beliefs and rituals are the main components of
any religion. Sociological definitions of religion take two main forms: substantive and
functional. Substantive definition defines religion as a belief and institution directed
towards deities or other super human beings such as ancestors or nature - spirits (Tylor,
1871). Functional definition of religion arose principally from Durkheim's rejection of the
Tylorian approach. According to Durkheim religion is a binding force and this balances the
growth of a society.
In history of society religion is based on the functional requirement of individuals. Functional
requirement is to have faith in something for ones own reassurance and confidence.
Evolutionists such as Tylor and Muller attempted to explain religion in terms of human
needs. Tylor saw it as a response to man's intellectual needs, Muller saw it as a means for
satisfying man's emotional needs. (Harlambo).
The religion has two forms. They are animism and naturalism. Animism is the belief in
spirits. Edward B Tylor believed this as the earliest form of religion. He argued that
animism derives from man's attempt to answer questions on the relationship between
life and death. Tylor suggested that religion, in the form of animism originated to satisfy
man's intellectual nature to meet his need to understand the events of death, dreams
and visions (Tylor, 1970). On the other hand, proponents of naturalism believed that
the forces of nature have some supernatural power. Contradicting Tylor's arguments,
Malinowski put forward that naturalism was the earliest form of religion. According to him,
naturalism arose from man's experience of nature, in particular the effect of nature
upon man's emotions. Nature contained surprises, terrors, marvels and miracles such
as volcanoes, thunder and lightning. Awed by the power and wonder of nature, the
primitive man transformed abstract forces into personal agents. The force of the wind
became the spirit of the wind, the power of the sun became the spirit of the sun
(Malinowski, 1954) Animism seeks the origin of religion in man's intellectual needs,
while naturalism seeks it in fulfillment of man's emotional needs. In the context with the
rural masses of India, one finds the peculiar blend of animism as well as naturalism
To some extent Durkheim in his book 'The Elementary Forms of Religion' has supported
the blend of natural power and the supernatural beliefs. He said, that all societies divided
the religious acts in to "the sacred" and the "the profane". Sacred things are considered to
be superior in dignity and power to profane (non-sacred) things. According to Durkheim,
religious beliefs are neither to fulfil intellectual needs nor emotional as suggested by Tylor
and Malinoski but religious beliefs and practices are needed for the survival of a man.
Religion in all forms and types has functional use in a man's life and therefore they were
always present.
The Latin 'Ritus' from which the term ritual is derived means 'custom', a notion which has
misled certain sociologists to believe that ritual was the routine of an organized religion.
There is, however, no denying that without ritual there cannot be an organized religion, but
this does not necessarily mean that all rituals are religious. There are many rituals which
exclusively have social character, not to speak of the magical and what we would like to
call the metaphysical rites, none of which can be confused with the religious ones. Certain
rites are as much a part of the daily routine of the individual and hence as much as eating,
drinking and the other odd chores of domestic life. Unless the necessitous is defined
strictly in physiological terms without any sociological admixture, rites cannot be placed in
the category of the extra-necessitous. And if the term is defined in this manner, not only
rites but several other practices too, will have to be included in the other category. The
popular distinction between the sacred and the profane, again does not seem to be a
sound basis for distinguishing rituals from ordinary practices. It is in fact the ritual 'touch'
which makes certain practices sacred, not that an act becomes ritual because it happens
to possess a sacred character. The objects and beliefs treated as sacred are sacred only
because they are endowed with a ritual-value.
Rituals are often understood as a form of symbolic action. Sometimes symbolic actions
differ from the ordinary ones. However, the distinctive characteristic of symbolic actions is
that they are not governed by the laws of logic which normally govern the other ordinary
All human beings believe in supernatural power That is there is a power beyond human-
power and knowledge. It is true even in the case of health and illness. People believe that
one is healthy and other is not because of the effect of some supernatural forces. There are
instances among the rural masses where illnesses are associated with God. Diseases like
small pox, chicken pox, measles and cholera are generally associated with a particular
God or Goddess or deities as well as the power of natural elements like certain kind of
wind pressure, sun light and Neem tree. In rural India, people believe that the health
problems arise due to the sins committed in last birth. They generally associate the
outbreak of epidemics with the non performance of certain rituals by the population.
Likewise, respondents of the study believed certain diseases can never be cured with any
amount of medical aids and they can only be cured through the help of define power which
can be aroused by offering, prayers, chanting of mantras' etc. they did mention to the
investigator that dreaded diseases like small-pox, plague, cholera have cures in divine
offerings and religious rituals.
In the back-drop of the above discussion, one can appreciate the importance of ritualistic
action with in the frame-reference of religion.
Using the above two measurements, we arrived at three levels of health-status (high,
moderate and low) and three levels of religiosity as presented in Table 1.
0 (0) 39
X =99.306, df= 4,
The Table 1 suggests the frequency distribution of respondents into various categories
of health status and religiosity. The table shows that religiosity and health-status are in
correspondence with each other. It means highly healthy respondents are the highly
religious persons. It shows, that the majority of respondents who are having good health
(ie. 69 out of 126) are termed as highly religious persons too. Likewise, those who have
scored low on scale of religiosity, have scored low on health scale too (i.e., 11/26).
However, it is worth noting that a very low percentage of the total respondents fall in the
category of low health status (i.e., only 26 respondents). Out of these 26 cases, only 11
are in the category of low health status. This finding suggests that most of the
respondents were very religious and therefore, they were following the traditional
practices. Some of the statistical results confirm the above finding. For example, the
Chi Square test score shows that, there is a significant relationship between (p < 0.01).
Religiosity and Health Status.
In addition to the above results, the coefficient of correlation also confirms (r = 0.6117) a
significant relationship. The respondents who are placed into the category of highly
religious are visiting the places of worship once in a day. They used to perform certain
daily routine practices as sacred functions or rituals, such as taking bath before going
to a temple, use of sandal mark on forehead, smearing of sacred ashes (made of burnt
cow dung cakes) on the fore head, carrying flowers and camphor sticks etc. Most often
highly religious respondents kept fast for a day once in a week along with certain kind of
restrictive diets on rest of the days. They normally consume vegetarian diet consisting of
items like card, fresh vegetables, unpolished rice, seasonal fruits, coconut, etc. Generally,
their food was served on banana-leaves Their practices suggested inoculation of certain
amount of discipline and regularity in their way of living which in turn was able to provide a
mechanism of maintaining good health. To some extent this assumption got confirmed
through the answer pattern of respondents. After having a discussion on ritualistic
practices as routine action, it would be useful if we can have a look at the offerings
performed by the respondents Each respondent was asked whether they are offering.
On response to the question, all respondent excepting two respondents performed
Table 2 Religious Rituals for common cure '
Kinds of Offering Number
Goat 192
Pongal 193
Money 115
Hair 4
Table 2 shows 205 respondents perform offerings in the form of animals, pongal, hair etc. for
common cures. It is interesting to observe that almost all the respondents performed either
for common cure or tangible benefits excepting 2 respondents who mentioned that they
did not believe in offerings. 83 Normally these offerings were made in the form of
promises at the time of sicknesses and as soon as the sick persons became healthy - the
promises of offering were fulfilled in front of the deities
Table 3 Socio Economic Statuses and Religiosity
LOW 2 3 5 (2.4)
MIDDLE 23 44 74 141
HIGH 1 8 52 61 (29.5)
Table 3 shows that there exists a positive relationship between socio economic status and
religiosity. A majority of respondents belonged to the middle SES Category. Among the
middle SES Category members a majority of the respondents are placed into highly
religious group. It is the same in the High SES Category. From Table 3 it is clear that the
religiosity is increasing with the socio economic status. The statistical analysis also confirms
it (contingency coefficient = 0.34). Table 3 suggests that the respondents who belonged to
high SES categories are highly religious and low SES respondents are less religious. It
shows that the religiosity increase with SES. It is interesting to suggest that some
meaningful relationship have been observed among the socio economic status of
respondents and the various forms of ritualistic practices of the respondents The variables
influencing the health behavior are caste affiliation, age, income, educational achievements,
and the practices etc. and health
LOW Middle High
Less 39
Moderate 19
High 0
Total 58
X = 183.42, df= 4,
p<0.01 p<0.01
Table 4 shows a positive relationship between caste status and the religious status. It
means that the respondents coming from low caste hierarchy have scored low in
ritualistic performances. Likewise, the higher caste respondents have shown high kind of
religiosity in ritual performance.
Respondents were categorized into three major categories of religiosity based on their
religious activities such as diet, fasting, offering etc. Likewise, the respondents were
composed of various caste groups based on their status placement in caste hierarchy
such as high caste, middle caste and low caste. In the Low Caste Category there were
39 respondents who were also placed into the less ritualistic category (see Col. 1 of
5.1) because they scored low 83 on the composite scale of ritualistic behavior. It
suggests that there was complete congruity among the respondents of low category
Respondents coming from low caste groups did not believe in rigidity of ritualistic action.
Hence, they scored low on ritualistic scale While 19 respondents were placed into
"moderately religious" category Table 4 further suggests that respondents belonging to
low caste category either have been placed into low religious or moderately religious
groups. It is interesting to point out that none of the respondents of the low caste
category were placed in ' highly religious' category. Table 4 further denotes that almost
all the high caste respondents were placed into highly religions or moderately religious
groups. (See Col. 3 and row 3) while none of the high caste respondents were found in
the less religious category (see Col.3). The statistical results i.e. the Chi square test
score also confirms the above hypothesis by showing significant relationship between
different caste categories and the level of religiosity (Chi Square = 183.42 p <0.01). The
differences are in the expected direction. The coefficient of correlation result ( =0.789)
also confirm of the hypothesis. After some probing, we learnt that many rituals were
commonly followed in a very religious manner for some tangible benefits. For
example, once Muthu a respondent from Mannadiyar caste was needed five thousand
rupees for purchasing a pair of cows For this he promised for offering his hairs to Lord
Murugan of Palani, in case of getting the money. After a week or so, has sent who was
working in another form had sent him the required amount. This kind of offerings were
common among the higher caste respondents As soon as Muthu got the money, he
want and offered his hair.
Table 5 Religiosity and Health among different Age Groups
I (%)
T (%)
3 4
6 17
2 7
3 7 14
8 8 22
4 9 21
34 (46
1 14
2 28
The Table 5 denotes that the' health status of different age categories of the
respondents. It shows some kind of uniform distribution in different levels of health
status. However, it is interesting to note that the respondents belonging to young age
categories are also keeping the health status matching to the old age categories. One
would assume that young persons would be healthier than the old persons. However,
this assumption is completely believed with the findings. Further, it is observed that
the older respondents were having more disciplined life than younger respondents. It is
quite possible that the differences in health status may be taking place because of the
disciplined routine life.
Table 6 Religiosity and Health Among Different Educational Categories
7 20 27
4 8 12
MODERATE 10 18 44 72
(47 7)
5 6 11 22
22 (66 7)
HIGH 1 51 52
2 17 19
11 (335)
9 (17) 16
3 (100)
Table 6 denotes the health status of different educational categories of the
respondents. It shows some kind of uniform distribution in different levels of health
status however; it is interesting to note that the respondents belonging to lower level
of education categories were also keeping the health matching to the highly
educated respondents. One would assume that highly educated persons should be
healthier than the less educated respondents. However, this assumption is not found
true. Further, it is observed that the less educated respondents were having more
disciplined life than highly educated respondents. The highly educated respondents
migrate to urban areas due to unemployment, because of that they pick up bad
habits such as smoking, consuming liquors etc. Therefore, the highly educated
respondents could not follow the routines regularly. Hence, the relationship between
health and religiosity among different educational categories is clearly seen.
Based on respondents' description of ritual practices, some characteristic of rituals
has been observed. They are:
(i) Rituals performed for tangible gains: Attainment of tangible objects like wealth
success, physical, power political gains etc. were the major motives related with
certain kind of distribution of money/gift etc. Once Murugan appeared in the
Secondary School Examination, he prayed to Lord Murugan that if he passes the
examination, he would offer a coconut. When he succeeded in the examination he
offered it. Likewise, rituals are performed for tangible benefits.
(ii) Supernatural ritual performances are part of the religious system for attaining
salvation in some form. Sometimes their effectiveness is presumed to depend
upon the "will” of a supernatural being; thus, when respondents speak of prayer or
supplication, they imply that the supernatural being who is addressed may fulfil the
wishes of the petitioner. In some rituals, however, the performance is automatically
effective provided that it is carried out according to certain prescriptions. For
example, in case of spread epidemic diseases like smallpox, measles (a variety of viral
disease). In the study villages, it was observed that, whenever a child is ill of
chicken pox the members of the family worshipping the child by saying The Goddess
has gone into the child.
(iii) Rituals as moral conduct: According to Durkheim in all modes of life, relating to
serious acts such as happiness, grief, sufferings, etc one is suppose to perform
certain prescribed rituals.
For example, thanks giving celebrations, death rites, funeral-procession etc. These
rituals are brought in practice form to inculcate some moral order and discipline.
Among the respondents of the study it was a common practice that whenever there
was a happy occasion like the birth of a child, or a marriage, they performed some
rituals to celebrate the happiness as a symbol of thanks giving Similarly, when a
person died, the villagers performed rituals on the second day which they call 'Paal
Uthuthal' (milk giving ceremony) This ceremony was performed to show certain
amount of respect towards the departed soul.
(iv) Transcendental aspect of rituals: Ritual imposes a transcendental obligation - an
obligation which does not stand or sanctions but enforces itself spontaneously Its
impact on human mind may be characterized in metaphysical terms as awesome, faith
or devotion in contradiction to the psychological 'appeal' of dynamic morality. Love,
compassion, charity and loyalty, the tenets of the dynamic morality are the universal
principles of human existence. In conforming to these, man simply obeys the law of
his nature; he will cease to be a human being if he refuses to abide by them.
An overview of some of the above it can be suggested that health is as much a socio-
cultural phenomenon as it is a biological explanation. Religious values such as deeds
of the past, attributing to sins committed by people and consequent of wraths of gods
and goddesses and treatment sought through magico- religious practices, are
indicators of the influence of our tradition and cultural life. With the spread of education,
exposure to mass media, urbanizing and industrializing influences resulting in
occupational and spatial mobility and economic well being, choice of people to accept
modern medicine over folk medicine has increased. Even villagers or tribal folks look
forward to modern medicine for relief from pain, sufferings or physical ailments.
Medicine, whether folk or modern has a dual nature. Irrespective of the technological
level of a society, people still would lend support to the physicians efforts with their
prayers and propitiation of gods and goddesses. This mix of scientific temper and faith
healing in medicine needs to be understood in the context and situation in which it
operates. It may be only making tall claims that modern medicine has stalked death. It
has only postponed death but at the same time, the scientific development has
increased the "at risk factor" for the health of man. In other words, the life span of man
has increased but his rate of becoming unhealthy has increased many fold. Indian
villagers in this modern world still wants to try out various systems of medicine and
when they feel dissatisfied with one, they are inclined to try their hand on another, till
they are forced to entrust themselves to the folk medicine which is close to their cultural
1. Banerji, D, Poverty, Class and health culture in India, New
Delhi, Prachi Prakashan, 1982.
2. Chan Ho, Suzanne, S.Y., Dietary beliefs in health and illness using grounded
theory, Social Science and Medicine, Vol 30, 1990.
3. Dumont, L., A South Indian Subcaste, Oxford, Oxford University Press, 1986.
Idler Ellen.L, Kasl Staus Laur V., Religion, disability, depression, and the timing of
death, AJS, Vol 47, No. 4, 1992 Jan.
4. Jarvis, George K; North Cott C, Religion and differences in Morbidity and
Mortality, Social Science and Medicine, Vol 25, No.7, 1987.
5. Vaux, Kenneth, Religion and Health, Preventive and Social Medicine, No. 5, 522-
536, 1976.
Full-text available
Social Network suggests that the decisions are not made in isolation but they are outcome of influence and confluence of social correlates (Srinivasan and Sharan 2005). They are products of consultations with other members of the community or the institutions. Many at times they happen mutually. Social exchange theory suggests that the interaction by the individuals is based on the mutual transactions. This also helps one to build bond with others. This is an important aspect of health. In general it is observed that the health seeking behaviour of individuals is influenced by other members of the community. When a need arises a person interacts with others and exchanges feelings and emotions required for health. The present paper explains the health seeking behaviour using systems approach. According to the framework, Health is a system in which there are three subsystems, which have many sub elements within each sub-system (Srinivasan and Sharan 2005). The three subsystems described are individual, community and health administration. Health of an individual is an outcome of the interaction among the subsystems. For example, an individual’s choices are influenced by his or her characteristics, the community members, the availability and accessibility of required services and so on. In addition it is further influenced by the interactions among the sub-elements of all the three subsystems. For an example, the sub-elements of individuals such as age, educational qualifications, income, occupation, awareness and so on influence one’s decisions on health. Likewise, the sub-elements in the community subsystem such as affiliation (caste), religion, friendship, and social climate and affinity affect health choices affect the one’s decisions. The sub-elements of the health administration sub-system, such as, the availability (provision of services and medicine, and timing), the accessibility (location, connectivity and communication), and attitude of personnel, are
It appears that religious beliefs and associated moral habits vitally affect health attitudes and behaviors. As a result of this causal relationship, health and disease indices vary in important ways according to the penetration of these beliefs into the daily fabric of the patient's life. Three religious dispositions, born of fundamental belief, anchor those moods and behaviors which profoundly influence health. As these three overarching themes purity of life, peace in existence, and belief in immortality are common to most religions, this paradigm might serve to organize further research on those belief characteristics which have health import. More work is needed to differentiate the relative importance of interacting variables where religion and health are considered; for this inquiry will produce insights with great significance for preventive medicine and public health. As the public response to health care possibilities and imperatives may depend on the penetration and pervasiveness of religious feelings and disciplines, the general culture will benefit from the discoveries of this investigation.
Dietary beliefs in health and illness using grounded theory
  • Chan Ho
  • S Y Suzanne
Chan Ho, Suzanne, S.Y., Dietary beliefs in health and illness using grounded theory, Social Science and Medicine, Vol 30, 1990.
Idler Ellen.L, Kasl Staus Laur V., Religion, disability, depression, and the timing of death
  • L Dumont
  • South Indian
  • Subcaste
Dumont, L., A South Indian Subcaste, Oxford, Oxford University Press, 1986. Idler Ellen.L, Kasl Staus Laur V., Religion, disability, depression, and the timing of death, AJS, Vol 47, No. 4, 1992 Jan.