Cultural and Social Factors Influencing Mortality Levels in Developing Countries Author(s): John C.

Caldwell Source: Annals of the American Academy of Political and Social Science, Vol. 510, World Population: Approaching the Year 2000, (Jul., 1990), pp. 44-59 Published by: Sage Publications, Inc. in association with the American Academy of Political and Social Science Stable URL: http://www.jstor.org/stable/1046793 Accessed: 14/08/2008 22:37
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=sage. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

http://www.jstor.org

ANNALS, AAPSS,510, July 1990

Culturaland Social Factors InfluencingMortalityLevels in Developing Countries

C. CALDWELL ByJOHN ABSTRACT: Recentanalysesof ThirdWorlddata,bothat the level of national or other large aggregatesand at that of individualsstudiedin sample surveys, have revealedthe surprisingfact thatsocial characteristics, such as the level of or or cultural such as ethnic group, characteristics, schooling fertility control, in are usually more influential determiningmortalitylevels than is access to medical services, income, or nutritional levels. Evidencefromthe UnitedStates at the beginning of the centurysuggests thatthis was not the case earlierin the West. This articleexamines the evidence, shows why developing countriesare evidence on how currentlyin an unusualsituation,andpresentsanthropological cultural,social, andbehavioralfactorsachieve theirimpact.An attemptis made to begin the constructionof a more generaltheoryof mortalitytransition.

John Caldwell receivedhis Ph.D. degree in demographyat theAustralianNational University, Canberra.He and his wife, Pat Caldwell,have researched populationchange via anthropological and demographic field-researchtechniquesin Thailand, Africa,India, and Malaysia,sub-Saharan Sri Lanka.From 1970 to 1988 he was head of theAustralianNational University's Departmentof Centre.He is authorof Theoryof Fertility Demographyand now is directorof its Health Transition Researchin South Decline; and coauthor of The Causes of DemographicChange:Experimental India. 44

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

45

mortalityin the West has given all the appearancesof supportinga commonsense and economic-deterministinterpretationof healthchange.Duringthatperiod, life expectancy in the most developed countries increased by more than 50 percent, from under 50 years to around 75 years. Nothing like this had happenedbefore in humanhistory.But no one was very because otherchanges of fundasurprised, mentalimportance had also occurred: with the full floweringof the Industrial Revolution, real incomes in the most economically advanced countries had multiplied almost tenfold over that period. This allowed people to be better fed and clad, permitted the construction of improved hospitals that the populace could increasingly afford to use, and provided the resourcesto treatdrinkingwaterandsewage. At the same time, the interrelated scientific revolutionfirst made safer medical procedures possible and ultimately produced sulfa drugs, antibiotics,new vaccines, and powerful insecticides. Material improvement and scientific advance seemed to have been the main engines driving down mortality,an interpretationthat this article will argue was not, for the West, far wrong. Furthermore, for most of this period, the interpretation was largely confined to the West, partly because most analysts lived there but largely because therewas little in the way of healthstatisticselsewhere.This fact was to cloud our interpretation of the forces behindthe globalmortalitytransition when it began to gathermomentumin the present century. Certainly,it was known that these advances were not equallyshared.As earlyas 1852 WilliamFarrhaddemonstrated major

OVER the last 100 years,the historyof

mortalitydifferentialsin Englandby socioeconomic class,' butthesewere assumedto reflect real differences in the means to bring minimum comfort and to pay for treatment and little else. Not everyone agreed that individualbehaviorplayed no significantrole in determiningthe level of mortality,as was shown by the aims and activities of the InfantWelfareMovement in English-speaking countries before WorldWarI or by the Maternaland Child Welfare Movements of the 1920s and 1930s.2A revisionist approachto the history of medicine has been developed in recent decades by ThomasMcKeown and colleagues,3 but the thrust of this work has been to emphasize the impact of economic change at the expense of scientific medicine.
NEW EVIDENCE FROMDEVELOPINGCOUNTRIES

A majorshift in ourinterpretation of the transition hasbeen madepossible mortality by a change of focus to the development problems of the Third World, where life expectancy in sub-SaharanAfrica is still below 50 yearsandin SouthAsia below 55 years. These are populationslargely lack1. William Farr, Vital Statistics: A Memorial Volume ofSelectionsfromtheReportsand Writings of William Farr (Metuchen,NJ:ScarecrowPress,1975). 2. Ellen Ross, "Mothers andthe Statein Britain, 1904-1914"(Paperdeliveredatthe Conferenceon the Historical Context and Consequencesof Declining Fertilityin Europe,CambridgeMeeting, July 1989); Philippa Mein Smith, "Infant Survival, the Infant Welfare Movement and Mothers' Behaviour, Australia with Reference to New Zealand, 1900-1945" (Ph.D. thesis, AustralianNational University, Canberra,1989). 3. Thomas McKeown, The Role of Medicine: Dream,Mirage orNemesis (London:Nuffield Hospitals Trust,1967).

46 ing adequate death registration systems and medical identificationof the cause of death,so adequatedatabankson mortality and social and economic conditions have been amassed by the United Nations and the WorldBank only in recentyears. This evidence shows that levels of income and healthservices areweak predictors of mortalitylevels and that social determinantsapparently play a majorrole in This articlesummadetermining mortality. rizes that evidence, attemptsto employ it to explain global mortalitytransition,and on how analyzes the availableinformation social factorsaffect deathrates. It has become increasingly clear that Third Worldnational mortalitylevels exhibit a very different pattern from what their income levels would imply. Some poor developing countries have largely escaped the Malthusianshackles. A 1985 Rockefeller Foundation Conference4selected for investigatory study four Third Worldsocieties thathad achieved low levels of mortality "at low cost": certainly, when comparedwith developed countries with similar mortality levels, at absolute low cost, but not always at relatively low cost if the measure is the proportionof nationalincome spenton healthandsocial services likely to assist the maintenance of health. The societies chosen-Sri Lanka; Kerala State, in southwest India; China; and Costa Rica-all had life expectancies in the 66- to 70-year range. In the case of the first threesocieties, this was at least 15 years higherthanthe averagefor countries with similarincomes, about3 years lower than Eastern Europe, and only 7 years below WesternEuropeandNorthAmerica. This level of mortalityhad been achieved
4. Scott B. Halstead,Julia A. Walsh, and KennethS. Warren, GoodHealth at Low Cost (New York: RockefellerFoundation,1985).

THE ANNALS OF THE AMERICANACADEMY

by these three societies with per capita incomes in the $300-400 range, or onefortieth that of WesternEurope and onefiftieth of North America.5Considerable attention was paid to the nature of their health and social services. Partof the explanationfor the high life expectancies probably lies in the socialservice net that Sri Lanka, China, and Costa Rica provide. It should be noted, however that Sri Lanka spends only 1.2 percent of its gross national product on health, slightly above the average for its income level, comparedwith 3.7 percent for Westernindustrialized countries,which in absolute terms, about 140 times spend, as much per person.6In termsof the number of inhabitants perphysician,Sri Lanka, with 7500 persons per doctor,or 15 times as many as in the West, is typical of its income level;7 Keralaprobablypresentsa similarpicture,althoughthe identification for statisticalpurposes of doctors in both India and China includes many persons whom other countrieswould exclude and so makes comparisons impossible. The conclusion is inescapable that neither income nor the levels of health services and interventionsare the primaryexplanation for the remarkable healthachievementsof these societies. The new data banks providedvaluable clues. Multivariate analysis allowed the level of a range of socioeconomic factors and health inputs in developing countries to be related to mortality levels. Flegg
5. Cf. WorldBank, World DevelopmentReport, 1988 (New York:OxfordUniversityPress, 1988), pp. 222-23, tab. 1. 6. Ibid.,pp. 266-67, tab.23. Thereareno figures for Kerala, which probably spends less, or China, while Costa Rica spends a proportionof the gross nationalproductin the range of the proportion spent countries. by the industrialized 7. Ibid., pp. 278-79, tab.29.

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

47

showed that the level of literacy was the best indicatorof low infant mortality,although the degree of equality in income and the level of medical care also played important roles.8 Caldwell demonstrated thatlow nationalmortality was most highly correlatedwith the proportionof females in school a generationearlier,and thatthe levels of family planningpracticeandmale school attendancewere also importantindicatorsof low mortality.9 Lower correlations were found with the ratio of doctors to populationand nutritionallevels, and a still lower correlationwith income levels. Recently, Rogers and Wofford have confirmed the prime role of literacy and the proportionof the populationworking outside agricultureand, of lesser importance, the safety of the water supply.'? Health inputs, as measuredby the ratio of physicians to population,showed a lower level of correlation, while nutrition was not found to be significant.The importanceof schooling a generation earlier lies in the fact that it determinesthe currentlevel of in parentaleducation,especially important the case of mothers.The markeddifferential in the survivalof childrenaccordingto the level of mother's education had been notedin Ghanain the 1960s by Gaisie" and in a range of LatinAmericancountriesin
8. A. T. Flegg, "Inequality of Income,Illiteracy andMedicalCareas Determinants of InfantMortality in UnderdevelopedCountries,"Population Studies, 36(3):441-58 (Nov. 1982). 9. John C. Caldwell, "Routesto Low Mortality in PoorCountries," RePopulationandDevelopment view, 12(2):179, tab. 3 (June1986). 10. Richard G. RogersandSharonWofford,"Life Expectancyin Less Developed Countries:Socioeconomic Development or Public Health?"Journal of Biosocial Science, 21:245-52 (1989). 11. S. K. Gaisie,Dynamicsof PopulationGrowth in Ghana, Ghana PopulationStudies no. 1 (Legon, Accra: University of Ghana, Demographic Unit, 1969).

the 1970s by Behm.Y2 The importanceof child survival for determining mortality levels lies in the fact that,in high-mortality of countries,typically at least one-quarter all birthsresult in deathsbefore 5 years of of the age, andbecause of the age structure half of all deaths in the population, society occur to personsunder5 years. Meanwhile, these macro observations were being increasingly supported by individual-leveldatacollected by national and subnationalrepresentativesurveys in the ThirdWorld.A majoropportunity was of surveysof presentedby the organization good scientific quality in 45 developing countries by the World Fertility Survey program in the decade after 1975. Two differentanalysesl3largely supported each other in their finding that parentaleducation is the most importantinfluence on child survival, with mother's schooling usually havingthe greaterimpact.Income, evidenced by father's occupation, is also important.Child mortality declines with every additionalyear of mother's education with no lower threshold,so that even one or two years of schooling in a rural school has some impact. Furthermore, social influences are of greaterimportance, as evidenced by wider differentialmortality between groups, for children aged 1-4 years thanfor infants,presumablybecause
12. Hugo Behm, Final Report on the Research Project on Infant and Childhood Mortality in the ThirdWorld(Paris:Comit6 international de cooperation dans les recherchesnationalesen demographie, 1983). 13. John C. Caldwell and Peter F. McDonald, "Influence of Maternal Education on InfantandChild Mortality:Levels and Causes,"in InternationalPopulation Conference,Manila, 1981, (Liege: International Union for the Scientific Study of Population, 1981), 2:79-96; J. N. Hobcraft,J. W. McDonald,and S. O. Rutstein,"SocioeconomicFactorsin Infantand Child Mortality: A Cross-National Comparison," PopulationStudies,38(2):193-223 (July 1984).

48 of the leveling effect on the latterof nearly universalbreast-feeding. Mensch, Lentzner, and Preston analyzed 15 surveysin Africa, Asia, andLatin America, only 6 of which were from the WorldFertilitySurvey program.They explored the impact of 12 groups of social, economic, and health-carevariables,both uncontrolled and then controlled for the influence of the othervariables.They concluded that the major influences on child survival were mother's education,ethnicity, and, largely in urban areas, father's education.14 The extent to which maternal education has been identifiedas a major- or even the child mortalmajor- factorin determining ity is astonishing,althougheven this finding merely provides clues to the forces at work ratherthan a simple answer. When two socioeconomically similar areas in Nigeria's Ekiti district were comparedin order to discover the mortalityimpact of different levels of health services, it was discovered that, even where there was no access to modem healthservices, mother's determinant of schooling was an important child survival.'5Researchin southwestNigeria, especially in Ibadan,for the Changing African Family Project, showed that the importanceof maternaleducation remained after controlling for the occupations of both motherand father,the urbanruraldivision and the residentiallocation within urbanareas, whether the marriage

THE ANNALS OF THE AMERICANACADEMY

was monogamous or polygynous, and whetherthe parentspracticedfamily planning or not.'6Mensch, Lentzner,and Preston concluded: Whenexamined by itself,anadditional yearof mother's reduces childmortality schooling by an average acrossour 15 countries of 6.8 perof countries in the cent,withthemajority falling of 5.0 to 9.0 percent. Afterallother varirange ablesareentered intothe estimation equation, the effectis still a reduction of 3.4 percent in
mortalityperyearof schooling.This latteris the "direct" effect of schooling andis biaseddownward as an estimateof the "total"effect by the inclusion of variables whose value is partly determined by mother'sschooling itself.17

Maternal education and child survival were the focus of two papers,the first by Clelandandvan Ginneken'8 andthe second alone. The Cleland latter concluded: by The most important features of the maternal education-childhood association mortality may
be summarizedthus: there is no threshold;the association is found in all major developing regions; the linkage is stronger in childhood than in infancy; only about half of the gross association can be accounted for by material advantages associated with education; reproductive risk factors play a minor intermediate role in the relationship; greaterequity of treatment between sons and daughtersis no partof the explanation; the association between mother's education and child mortality is slightly greaterthan for father'seducationand mortality.19

16. John C. Caldwell, "Educationas a Factorin Mortality Decline: An Examination of Nigerian 14. United Nations, Socio-Economic Differen- Data,"Population Studies,33(3):395-413(Nov. 1979). tials in Child Mortalityin Developing Countries,by 17. United Nations, Socio-Economic DifferenBarbara andSamuelPreston, tials, p. 287. Mensch,Harold Lentzner, ST/ESA/SER.A/97 18. JohnClelandandJeroenvan Ginneken,"Ma(New York:UnitedNations,1985). 15. I. O. Orubuloyeand John C. Caldwell, "The ternal Educationand Child Survival in Developing Impact of Public Health Services on Mortality:A Countries:The Search for Pathways of Influence," Study of MortalityDifferentials in a RuralArea of Social Science andMedicine,27(12):1357-68 (1988). 19. JohnCleland,"Maternal Nigeria," Population Studies, 29(2):259-72 (July Education andChild Survival: Further Evidence and Explanations,"in 1975); Caldwell, "Routesto Low Mortality."

MORTALITY LEVELSIN DEVELOPINGCOUNTRIES

49

ceedings of Workshop,Canberra, May 1989, ed. John C. Caldwell et al. (Canberra: Australian National University, Health Transition Centre,1990). 20. John C. Caldwell, "MassEducationas a Determinant of Mortality Decline" (CASID Lecture, Michigan State University,25 Oct. 1988), reprinted in Selected Readings in Cultural,Social and Bhavioural Determinantsof Health, ed. John C. Caldwell andGigi Santow(Canberra: Australian National UniCentre,1989),pp. 103-11. versity,HealthTransition 21. Julie DaVanzo, William P. Butz, and JeanPierre Habicht, "How Biological and Behavioural

peoples of the West African savanna in ruralMali.22 Mensch, Lentzner,and Preston examined60 ethnic groupsin 11 countriesof Africa,Asia, andLatinAmericaand almost invariablyfound significant ethnic differentialsin child survivalin each country, with the mortalitylevel in one group Thereis as close a correlation betweenchild sometimes being twice or more that of survivaland generallevels of [female]edu- another.They noted that Chinese populacation in a communityas there is between tions in SoutheastAsia arecharacterized by childsurvival andmaternal An edu- unusuallylow mortality.23 education. cated woman in acounOne aspect of the impacton child mormayfeelmore deprived womenareeducated than tality of different cultural attitudes and trywheremostother in one wherethey are not; nevertheless, her practicesis thatof preferencefor sons over children standa muchgreater chanceof surdaughters. This preference, where it is vival.If we takethesetwofactors the together, found, almostcertainlykeeps child mortalcontrasts areenormous. In LatinAmerica, the thanit mightotherwisebe in that deathrateamongthe children of uneducated ity higher Peruvian womenis almost 7 timesgreater than the preferential treatmentis unlikely to force male mortalitydown by as great an Venezuelan women with seven among years additional increment as it unnecessarily of education. InAsia,themortality chilamong female mortality high. The World drenof uneducated women is almost keeps Nepalese 15 timesgreater thanit is among thoseof Ma- Fertility Surveys provided for the first laysianwomenwith seven or moreyearsof time, by the use of a life-historyapproach, schooling.20 accuratedataon child mortalsubstantially and sex. This showed that in One persistent, but underresearched, ity by age the sensitive age range of 1-4 years there finding is that there are major ethnic or is excess female over male mortality cultural differentials in mortality, especially child mortality, even in the same throughout nearly all North Africa, the countryandwith the same access to health Middle East, South Asia, and East Asia, services- differencesthatsurvivecontrol- with greaterdiversityin SoutheastAsia and ling for income and education. This has Latin America, and little or no additional been shown between Chinese and Malays danger for females only in sub-Saharan Data on differentialsex mortality in Malaysia2' and between the different Africa.24 from the Indian Sample RegistrationSurWhat We Know about Health Transition:The ProDiarrhoealDisvey and the International an International
Influenceson Mortalityin MalaysiaVaryduringthe FirstYearof Life,"PopulationStudies,37(3):381-402 (Nov. 1983). 22. Allan G. Hill, ed., Population Health and Nutrition in the Sahel: Issues in the Welfareof Selected WestAfrican Communities(London: Kegan Paul International, 1985). 23. United Nations, Socio-Economic Differentials, pp. 77-111. 24. Shea O. Rutstein,"Infantand Child Mortality: Levels, Trendsand DemographicDifferentials," Studies:Cross-National Comparative Summaries,no.

The point thatmay not have been sufficiently stressed is that education has two separatebut multiplicativeimpacts,one on individualswhose behavioris changedrelative to their society and one thatchanges the whole society. An examinationof the WorldFertilitySurveys concluded:

50 eases ResearchCentre'sBangladeshpopulation laboratoryin the Matlab district demonstrate how culturally specific beFemalemortalhaviorcan affect mortality. not above that of males is ity duringthe first of when life, year breast-feedingprovides and protectiveantibodequal nourishment is it ies; relatively high in the 1- to 4-year age range,when thatprotectionhas dwindled butchildrenarestill highly dependent on others;it falls towardparitybetween 5 years and marriageas girls become more capable of fending for themselves; and it rises above that of males again duringthe reproductive years,largelybecauseof high maternal mortality in the poor obstetric conditions of much of South Asia. Thereafterit falls below male levels.5 There is convincing evidence that the achievementof a small family,or even the intentionof havingone by employingbirth control,is associatedwith declines in child mortality.There is a correlationbetween nationallevels of child mortalityand family planning practice that compares only with thatbetween maternaleducationand The Nigerian segment of the ethnicity.26 African Changing Family Projectfound in Ibadancity child-mortality levels thatwere far lower among those women who had achieved relatively small families than The One-peramong those who had not.27
43, rev. ed. (London: World Fertility Survey, Dec. 1984); Pat Caldwell and John C. Caldwell, "Where There Is a NarrowerGap between Female and Male Situations:Lessons from South Indiaand Sri Lanka" (Paper delivered at the Social Science Research CouncilWorkshop on GenderDifferentialsin Mortality in South Asia, Dhaka,Bangladesh,Jan. 1987). 25. Caldwell and Caldwell, "WhereThere Is a NarrowerGap." 26. Caldwell,"Routesto Low Mortality," p. 179, tab. 3. 27. John C. Caldwell and Pat Caldwell, "The Achieved Small Family: EarlyFertilityTransitionin an African City,"Studies in Family Planning, 9(1):218, app. B (Jan. 1978).

THE ANNALS OF THE AMERICANACADEMY

ThousandSurvey of Chinafound extraordinarilylow mortality amongonly children whose parents had completed the documentationopting for that status.28 The interrelationshere are complex and are discussed in the next section. I continued furtherwith the approach adoptedin the "GoodHealthat Low Cost" conference.29 A comparisonof the mortality and per capita income rankingsof the 99 ThirdWorldcountriesreportedfully by the WorldBank because their populations exceed 1 million showed that, in terms of their income, 11 did exceptionally well with regardto health,being 25 to 62 places higher in their health rankingsthan their incomes would have predicted,while another11 did exceptionallybadly,falling 25 to 70 places below prediction.Armedwith this informationand that from correlation analysis, together with anthropological, on sociological, and historicalinformation the societies thathadbeen most successful in driving down mortalitywithin their income constraints, the studycame to a numberof conclusions.Parental educationis of great importance,especially that of mothers. So is the controlof fertilityor even the attemptto control it. Female autonomyis important,and its relative lack was the main reasonthat 9 of the 11 countriesless successful in convertingtheirincomes into low mortalitywere found in the western branchof Islam stretchingfromSenegal to Iran. Grass-rootsradicalism, egalitarianin both ism, anddemocracywere important a successful demand for creating popular health and educationalservices and ensuring that they worked. Neither female autonomy nor radicalismhas as yet been as successfully researched as education,
28. John C. Caldwell and K. Srinivasan,"New Data on NuptialityandFertilityin China," Population and DevelopmentReview, 10(1):71-79 (Mar.1984). 29. Caldwell, "Routesto Low Mortality."

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

51

partly because there are no simple measures equivalent to years of schooling.30 Clearly,as is discussedin the next section, these characteristicsare interrelated.The studywent fartowardsuggestingthatsocieties are largely prisonersof theircultures and historiesand that the roots of contemporaryhealthsuccesses lie farbackin those histories. The exceptions were the successes achievedby Communistrevolutions in ChinaandVietnamand,less certainlybecause it earlierwas markedby some of the othercharacteristics-by Cuba. Wherethe greatestsuccesses over mortality have been gained, this achievement has been the productof an interactionbetween certainculturalandsocial characteristics on the one hand and the easy accessibility of basic modem healthservices on the other. In spite of the fact that parental educationand the practiceof fertilitycontrol correlate so much more highly with mortalitylevels than do medical interventions in the contemporary ThirdWorld,the evidence stronglysuggests that alone they cannot make dramaticreductionsin mortality levels. They may, in fact, correlate morehighly becausehealthinvestmenthas been runningaheadof social investmentin termsof the optimummix. Sri Lankahad experiencedmassive social change by the 1920s. The 1921 census had found 56 percentof males and 21 percent of females to be literate, a level that Pakistanwas not to reach for anotherhalf century.Yet life expectancywas little over 30 years.3'It was the provision of health
30. John C. Caldwell and Pat Caldwell, "Women's Position and Child Mortalityand Morbidityin Position and DeLDCs,"in Conferenceon Women's mographic Change in the Course of Development, Asker (Oslo) 1988 (Liege: International Union for the Scientific Study of Population,1988), pp. 213-36. 31. Caldwell,"Routesto Low Mortality"; T. Nadarajah,"Trendsand Differentials in Mortality,"in Population of Sri Lanka (Bangkok: Economic and

services, firstin urbanareasandthen,from 1945 onward, rapidly in rural areas, that allowed the subsequent dramatic fall in deathrates.Yetearlierthe countryhadbeen highly sensitiveto the need to combatsickness and possessed one of the most extensive and developed systems of traditional medicinein the world.Traditional medical systems may provide solace and reduce pain and even symptoms in chronic or other conditions, but the evidence seems clear that modem medicine is needed to drive down mortalityrates. Francehad reducedits fertilitylevel to the equivalentof 3.5 birthsper woman by 1850, but its life expectancywas only 39 years.32Three societies that had experienced a great deal of social change-Sri Lanka, Kerala, and Costa Rica-enjoyed periods of intensive activity when health services were spreadmuch morewidely to the ruralpopulations and the urbanpoor during the years 1946-53, 1956-71, and 1970-80, respectively;in each case mortality fell muchmorerapidlythanin earlieror lateryears.33 Therearealso societies where lack of specific types of social change, often female autonomy or female education, meansthatmajorinfusionsof modem healthservices do not achieve theiranticipated impact. By 1980 Libya employed more doctorsper capita thanJapanor Ireland and was reaching the levels of the UnitedKingdomandNew Zealand,butlife expectancytherewas 16 years shorterand the infant mortality rate seven times as high. There is, then, some kind of symbiosis between social change and modernmediSocial Commission for Asia and the Pacific, 1976), p. 148. 32. NathanKeyfitz, World Population:An Analysis of Vital Data (Chicago: University of Chicago Press, 1968). 33. Caldwell,"Routesto Low Mortality," p. 181.

52 cine, the lattermeasuredmoreby its accessibility to a wide population than by its level of technology.A comparisonof two socioeconomically similar populations in Nigeria, one with access to a hospital and doctors and the other isolated from such interventions,suggests thatthe gain in life expectancy- equivalentto the moreeasily measuredchanges in child mortality- was 20 percentwhen the sole intervention was to access health facilities for easy adequate illiteratemothers, 33 percentwhen it was educationwithout health facilities, and 87 percentwith both.34 Clelandandvan Ginnekensummarized datafrom a wide rangeof countriesshowing that the use of modem health services increased with duration of education.35 They believed that most evidence showed the interaction between education and health services to be less spectacularthan that found in Nigeria.36 They reachedthe conclusion that aboutone-half of the very great differentialsfound across the Third World in child survival by education of mother are probably explained by "economic advantagesassociated with education (income, water and latrine facilities, clothing, housing quality, etc.)."37They were more cautiousabout how the "pure" impact of mother's education was to be divided between interactionwith modem medicine and behavioraland care factors that prevent childrenfrom becoming sick or having an accidentin the firstplace, but they emphasizedthatbothwere likely to be important. Income also interacts with health-serviceprovision, and this interac34. Ibid., p. 204; cf. Orubuloye and Caldwell, "Impactof Public HealthServices." 35. Clelandandvan Ginneken,"Maternal Education,"pp. 1361-62. 36. Ibid., pp. 1362-63. 37. Ibid., p. 1360.

THE ANNALS OF THE AMERICANACADEMY

tion is especially strongwhen thereis little attemptto provide a free healthservice. A rangeof researchers attribute most of the pureeffect of maternal educationto the betteruse of modernhealth services, but a an imporsignificantnumberalso attribute tant role to family health managementindependent of curative services.38 In a Nigerianvillage thatwas so far from modem health services that very few children had ever been taken to doctors or nurses, motherswith some schooling experienced only one-third the child loss of mothers with no schooling. Only some of this can be explained by greater use of modem pharmaceuticals such as the malaria suppressants broughtby a cyclist who ran an itinerantpharmacyservice.39It might also be noted that the skills in healthmanagementthatcan preventchildrenfrombecomingsick or dyingin the absenceof modem health services are the same ones that allow healthservices to be exploited more successfully.Further convincing evidence of culturaland social differentialsin child mortalityin the pre-moder-medicine era has been providedby researchamong societies in ruralMali where the modem era has not yet begun. The substantialdifference in child mortalitybetween adjacent culturalgroupswas explainedby different In contemporary sostyles of child care.40 cieties, some familiesaremuchmoreprone to experiencesickness andto lose children than others, as was shown forty years ago
38. For detailed references, see Caldwell and Caldwell, "Women'sPosition,"pp. 222-23. 39. Orubuloyeand Caldwell, "Impactof Public HealthServices,"p. 268. 40. KatherineHilderbrand et al., "ChildMortality and Careof Childrenin RuralMali"(Paperdelivered at the National Institutefor ResearchAdvancement and IUSSP Seminar on Social and Biological Correlatesof Mortality, Tokyo, 24-27 Nov. 1984).

MORTALITY LEVELSIN DEVELOPINGCOUNTRIES

53

and for Newcastle-upon-Tyne,England,41 in India.42 recently

INTERPRETING THE EVIDENCE A THEORYOF TO FORMULATE HEALTHTRANSITION

The first propositionis that there have in alwaysbeen socioeconomicdifferentials mortalitylevels and thatthey predatedthe impactof modernmedicine.This situation was partly a function of income and the abilityto eatbetterandenjoy othermaterial But the evicomforts, as Malthusnoted.43 dence on social differentials by ethnic group, and of greaterpropensityfor child loss in some householdsthan others, even Enin ratherhomogeneous contemporary urban areas or Indian villages, sugglish were also imgests thatsocial differentials will ever It is that research portant. unlikely populations identifypre-modern-medicine with no social differentials in mortality, especially child mortality,but it is highly probable that the differentials will be smaller than those in the era of modem medicine. It should be noted that this era did not suddenlybegin. Moreover,modem
41. J. Spence et al., A Thousand Families in An Approach to the Study of Newcastle-upon-Tyne: Health and Illness in Children (New York: Oxford UniversityPress, 1954). 42. Monica Das Gupta,"Death Clustering,MaternalEducationand the Determinants of Child Mortalityin RuralPunjab,India,"in WhatWeKnowabout Health Transition, ed. Caldwell et al. 43. Cf. John C. Caldwell, "Family Change and Change:The Reversalof theVeneration Demographic Flow," in Dynamics of Population and Family Welfare 1987, ed. K. SrinivasanandS. Mukerji(Bombay: Himalaya,1988), pp. 71-96; JohnC. CaldwellandPat Caldwell, "FamilySystems:TheirViabilityandVulTransactions nerability:A Study of Intergenerational and Their DemographicImplications"(Paper delivered at IUSSP Seminar on ChangingFamily Structures and Life Courses in LDC's, East-WestPopulation Institute,Honolulu,HI, 5-7 Jan. 1987).

medicine has become ever more effective, so thatsocial differentialsin mortalityarising out of interactionwith modern medicine are likely to have increased in the presentcentury. The second proposition is that a substantialpart,probablythe majority,of the explanationfor social differentialsin morThirdWorldlies tality in the contemporary in the interactionwith modern medicine. Evidencefor the mechanicsof this interaction is presentedat the end of this article. The interfacebetween society and modem of medicine is broaderthanthe proponents scientific medicineusuallylike to admit.It includes not only doctors, nurses, midwives, and pharmacistsbut also pharmaceuticals distributed through traditional markets, by wandering untrained salesboth men, and,on a massive scale, through traditionalmedical practitionersand nontraditional untrained practitioners or quacks. This informal system helps to changebeliefs andpracticeswith regardto illness and its treatmentand increasingly acts as a referral system to the more formal health sector. It is also probablethat this uncontrolledspread of modern medicine saves more lives ratherthan causes additional deaths, although the whole matter has hardlybeen researchedat all. This informalsectoris the only channelof modem medicine to much of ruralSouth Asia and sub-SaharanAfrica and almost certainly plays a role in the continuing decline in mortality in both regions. The impact of modem medicine in the formal sector is a functionless of its scientific levels thanof its accessibility throughrural clinics and nationalhealth schemes reducingthe cost to the patient. The breakthroughperiods in reducing mortality levels in different Third World countries have been associated with the democratization of services,

54 not with an increasein the qualityof medical technology. The thirdpropositionis thatthe various social mechanismsidentified as playing a role in reducingmortalityare reallydifferent facets of the same phenomenon,which might be called social modernization,or the riseof individualism orWesternization. It is really somethingbroaderand in many of the transiways is the social counterpart tion from subsistence production to the marketeconomy. It is the move toward a system where individualshave optionsand can exercise choices - andrealizethatthey can do so and act on that realization.The education, findingswith regardto maternal female autonomy,and grass-rootsradicalism are all part of this picture. It is also why sudden social shocks can accelerate processes,as theFrenchRevdemographic olution and the Japanesedefeat in World WarII did in the case of the fertilitydecline and as the Chinese Revolution did with regard to mortalityeven if its stated aim was farfrom the promotionof individualism. It is the dismantling of the subsistenceproduction organization and the control and belief systems that was necessary to ensure survival. Underlying these changes were profound economic changes. It was economic growththatproducedthe full marketeconomy and ultimately allowed individualsa degree of independencefrom the unified family economy.Nevertheless,in termsof demographicbehavior,the shifts in belief In the areaof systems were very important. health the important changes were toward a belief that sickness and death were the resultof nondivineand nonmagicalforces of this world,thattherewas somethingthat could be done about them in the form of either carefulbehavioror seeking the best help, andeventuallythatmodernmedicine was usually the most effective help that

THE ANNALS OF THE AMERICANACADEMY

could be obtained.In a study area in rural India,we called the processthe secularization of health behavior.44 That secularization does not necessarilyinvolve a diminution of religion, but it does involve its retreat from intervention in causing the everyday disasters of this world. In the Indianvillage the decline of the so-called little tradition andits village goddesses and of evil profusion spirits,in the face of the tradition of mainstreamHinduism, great associatedwith literacy,courts,and cities, is an example of this, as was the Puritan movement in England, which eventually moved ordinarypeople toward the view that most earthlyphenomenain their dayof material cause to-daylives were a matter andeffect andthatto thinkotherwisecould be blasphemous. It might be noted that earlier behaviorwas not irrationalin that much less could at that time be done to avertsickness and death. The West experienced these changes first because of unparalleled economic growthfrom at least the sixteenthcentury. It was eventually to export some of this growthas the worldmoved towarda global economy,but in the process it exportedits behavioralbeliefs and social attitudesand ultimatelyits medical technology and accompanying health philosophy. This exportwas achieved by colonial administrators, missionaries, the media, and, most powerfully,by the moderneducationsystems that are laden with Western,market, so-called rationalvalues with regardto behavior and family relationshipsand systems. These conduitswere so effective because they were hardlyconscious of their proselytizingrolebutusuallybelieved they
44. JohnC. Caldwell,P. H. Reddy,andPat Caldwell, "The Social Componentof MortalityDecline: An Investigationin South IndiaEmployingAlternative Methodologies," PopulationStudies, 37(2):185205 (July 1983).

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

55

were propagatingeither objective truthor objectively desirablebehavior.The impact of such ideological exports has probably played a central role in the near-global fertilitytransitionthat is occurring.45 There is compelling evidence that the impactof maternaleducationon child survival is not merely a case of learningmore evidence abouthealth.The most important is that it occurs everywhere: in good schools with good teacherswho do teach about health and in poor schools with underqualifiedteacherswho devote no time to the subject,as well as in every partof the Third World. Even stronger evidence is providedby the linearimpactof education so that even a little elementaryschooling has a proportional impact. Clearly,we are the witnessing generalimpactof ideas, ideologies, and behavioral models. In rural foundthatmothBangladesh,Lindenbaum ers who had been to school were cleaner andraisedtheirchildrenmorehygienically andcarefully,not becausethey hadlearned thatthis would save the children'slives but because they assumedthatthose with eduIn cationbehaved in such a superior way.46 a south Indian rural area, we found that mothers with schooling associated themselves and their schooling much more institutions- indeclosely with "modern" pendence and five-year plans as well as health centers and the case for using
45. JohnC. Caldwell, Theoryof FertilityDecline (London:Academic Press, 1982), esp. chap. 9, "The Failureof Theories of Social and Economic Change to ExplainDemographicChange:Puzzles of Modernization or Westernization," pp. 269-300. 46. Shirley Lindenbaum,Manisha Chakraborty, and MohammedElias, "The Influence of Maternal Education on Infant and Child Mortality in Bangladesh" (Reportfor the International Centrefor DiarrhoealDisease Research, Bangladesh, 1983), reprintedin Selected Readings, ed. Caldwell and Santow, pp. 112-31.

them-than did illiteratemotherswho felt that they were not partof this new world. The educatedalso felt this aboutthe uneducated,thus reinforcingthe latter'smental set.47 A corollaryof this argumentis thatmaternaleducationis likely to producemuch greater differentialsin child mortality in ThirdWorldthan in the the contemporary West,even the historicalWest,becausethe behavioralsystem had almarket-attuned readyevolved in the West.This bald statement might be modified by noting that education in the West did accelerate the working classes' adoptionof middle-class values and that "rationalindividualistic" behavior has continued to develop in the West. Prestonhas producedevidence from the U.S. census of 1900 to show that the gap between educatedprofessionalclasses and the rest of the society in child survival was much smaller than in the contemporaryThirdWorldandevidence from Baltimore in 1915 to show only small differentials between literateand illiteratemothers once father'sincomehadbeen controlled.48 He arguedthat this was because the level of health ignoranceof the middle class in America of that time was high and closer to thatof the workingclass thanis the case in the contemporary ThirdWorld.I subsequently contested this view, partly on the basis of a social-historicalstudy of health behavior in nineteenth-century Australia,
47. JohnC. Caldwell, P. H. Reddy,and Pat Caldwell, The Causes of Demographic Change: ExperimentalResearchin SouthIndia (Madison:University of Wisconsin Press, 1988), esp. chap. 6, pp. 132-60 and chap. 7, pp. 161-86. 48. Samuel H. Preston,"Resources,Knowledge and Child Mortality:A Comparison of the U.S. in the Late NineteenthCenturyand Developing CountriesToday,"in International PopulationConference, Florence 1985 (Liege: InternationalUnion for the Scientific Studyof Population,1985), 4:373-86.

56 and maintainedthat the lesser differential was due to the fact that modernmedicine was much less developed and had relaI now believe thatmy tively little to offer.49 was interpretation only partof the explanation and that the other part was that, as Westerneducation'smessagewas so much closer to the Westernbehavioralpatternof the time, its impact on changing attitudes and behaviorwith healthimplicationswas much less than in the contemporary Third World. It might be noted that Prestonand Ewbank have produced a study showing U.S. child-mortalityrates by social class widening between 1895 and1925, "consistent,"they argue"with the fasteradoption of behavioral innovations by the upper class groups."50 Among the behavioralinnovations that they documentis the growing resort to modern and increasinglyeffective medicine by professionalclasses. One furtherpoint should be made with regard to education. All contemporary Third World data show a significant impact on child mortalityfrom fathers'education as well as that of mothersand that much of this effect survivescontrollingfor income. Discussionhas centeredundulyon the maternaleffect, even though the existence of such an effect is good evidence that education probably affects both a father's attitudeand behaviorwith regard to his children's health and also his relasectionshipwith his wife, with a resultant ondaryimpacton his children'shealthand treatment. There is a related but distinct matter with regardto children'shealth.Thatis the
49. Caldwell, "Routes,"p. 206. 50. Douglas C. Ewbankand Samuel H. Preston, "PersonalHealthBehaviourandthe Decline in Infant and Child Mortality:The United States, 1900-1930," in WhatWeKnow aboutHealth Transition, ed. Caldwell et al.

THE ANNALS OFTHE AMERICANACADEMY

matterof the intrafamilialemotional and resource-allocation priorities. These change as the market develops and with Westernizationbut are better treatedas a separatestrandin social changelikely to be accelerated or retardedaccording to the natureof family structures.I have called the intrafamilialflow of resourceswealth flows and the change that directs more of them toward children than parentsor fathers the reversal of the intergenerational wealth flows to a downwarddirection.51 The fourthpropositionis thatchild mortality will fall more rapidly as the intergenerationalwealth flow turnsdownward. This almost inevitablyhappensas fertility declines. Indeed, not only does parental concern for child survivalincrease,as has happenedin contemporaryChina, but so does community and national interest in encouragingparentsto care for these increasingly rare and precious commodities, as happenedin the case of the infantwelfaremovementin theWestfromaround 1900 as the full extentof the recentfertility decline became clear.Families are willing to spend more effort and a greaterproportion of income on child care and survival. The situation is even more complex than this because there is a correlationin the Third World between the level of family planningpractice and child survival even before fertility decisively declines. The reason appearsto be that the wealth flow has begun to turn;the families are already placingmoreemphasison childrenrelative to the old, areplanningfor theirfuture,are finding that the adequateallocation of resources to each child for the successes of those plans can be attainedonly with fewer children, and are trying harderto ensure theirsurvival.
51. Caldwell, Theoryof FertilityDecline.

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

57

ELABORATIONS

Two modifications need to be introduced to this picture of social and family change. The first is that the rate of family changedependsto a considerableextenton its preexisting structure. In sub-Saharan Africa, partly because of the lineage system and partlybecause of widespreadpolygyny, wives usually have separatebudgets from those of their husbandsand are themselves responsible for many of the This resources needed by their children.52 a of mothers deal autonomy gives great with regard to health decisions affecting both their own and their children'shealth, but it often severely limits the resources available. In these circumstances, a of the spousalemotionaland strengthening economic bond, as well as any movement toward monogamy, is likely to accelerate decline. Similarlyin South child-mortality Asia and elsewhere, nuclear-familyresidence in contrastto extended-familyresidence is likely to give the young mother greater control over her children's health treatment.In south India, education can produce a degree of emotional nucleation even within the extended family and can give a mother greatercontrol over health decisions affecting her children. The second modificationis thata strong culturaltraditionlimiting women's autonomy, especially when reinforcedby a religion that regardsthe seclusion of women as a prime moral objective, can have a deleteriouseffect both on female healthin general and on all child health because of

52. John C. Caldwell and Pat Caldwell, "The Cultural Context of High Fertility in Sub-Saharan Africa,"Populationand DevelopmentReview,13(3): 409-37 (Sept. 1987); JohnC. Caldwell,PatCaldwell, and Pat Quiggin, "The Social Context of AIDS in 53. T. Valkonen,"SocialInequalityin the Face of PlenarAfrica,"Population and DevelopmentReview, 15(2) Death,"in EuropeanPopulationConference: Union for the Scientific Studyof ies, ed. International (June 1989).

the limitationin mothers'takingquick and effective action. This is the major reason why the Arab world does conspicuously badly relative to income in attaininglow mortality.Such traditionscan limit the educationof women and can limit the health effectivenessof thateducation.This is why in the WorldFertilitySurvey programthe differentialsin child survival by mother's educationwere so small in Bangladeshand why in Syria and Jordan, although they who had wereconsiderable betweenmothers never been to school and those who hadpossibly a culturalor ethnic effect-they were very small by durationof schooling. The final propositionis thatcultural,social, andbehavioralfactorshave an impact bothon an individual'smortalityandon the mortalityof an individual'sdependents.So muchanalysishasbeen carriedout on child survival because demographers' techniquesfor estimatingmortalitylevels from most ThirdWorlddata are much better at the youngest ages. This has also allowed the specific study of the impact of the mother'ssocial characteristics, because of the particularly importantrole she usually of young children.It plays in the treatment would be unwise, however, to believe that parentsplayed the sole role in ensuringthe survivalof children.In much of the Third and siblings play imWorld,grandparents roles. The portant improved relative survival chance of girls after5 years of age in South Asia shows that increasingly children play a role in their own survival and presumably, then, that their own social characteristics are increasinglyimportant. There are data now for Europe showing that mortality rates for adult males are much lower among the more educated,53

58 evidence probablyof a firmerdecision to controltheirown lives andfate. The determinantsof dependentchildren'smortality are a point of prime importancein highmortality societies, for where life expectancy is below 50 years andthe population is growing at 3 percent per annum, as in sub-Saharan Africa, the majorityof deaths are likely to be to personsunder5 years of age. Witha life expectancyof 60 yearsand a growth rate of 2 percent,however, as is now found in Southeast Asia, that proand portion drops to aroundone-quarter; with a life expectancy of 75 years and a growth rate of 0.5 percent,as now characterizes the West, the proportionfalls to one-fortieth.Hence, as mortalityfalls, the emphasis on health behavior shifts from parenthoodto how the middle-aged look afterthemselves.
THE TRANSLATIONOF BEHAVIORINTO SURVIVAL

THE ANNALS OF THE AMERICANACADEMY

This section documentsthe fact thatbehavior,especially mother'sbehavior,can, in ThirdWorldsocieties, be translated into lower child mortality.It focuses on rural south India,with a life expectancy around 50 years, and Sri Lanka,nearing70 years, areas where I have undertakenanthropological studies of demographicbehavior.54 The term "health management" describes behavior that prevents sickness
Associationfor PopulaPopulationand the European tion Studies,for the FinnishCentralStatisticalOffice Statistical Office, 1987),pp. 201-61. (Helsinki:Central 54. Caldwell, Reddy, and Caldwell, Causes of DemographicChange; JohnC. Caldwellet al., "Sensitization to Illness and the Risk of Death:An Explanation for Sri Lanka'sApproachto Good Health for All," Social Science and Medicine, 28(4):365-79 Social and (1989); John C. Caldwellet al., "Cultural, BehaviouralDeterminants of HealthandTheirMechanisms:A Reporton RelatedResearchPrograms," in WhatWeKnowaboutHealth Transition, ed. Caldwell et al.

from occurringor limits the damage once it does occur.Greaterfemale autonomyor educationincreasesa woman's capacityin health managementin two ways: first, by and selfgiving her greaterdetermination confidence and, second, by reducing the family and other constraintsplaced upon her. In traditional society, child care is Galalel Din often a diffusedresponsibility. showedhow in a Sudanesevillage children were rathercasually looked after by the whole village, as well as by theirsiblings, but, as mothers became more educated, they took greatercontrolandresponsibility themselves.55A research program in a north Indianvillage showed how women in semiseclusion had little confidence in their ability to identify sickness or to take the appropriate In ruralsouthIndia, steps.56 we foundthatmoreeducatedmothersgave greater emphasis to cleanliness, hygiene, nutrition,and the need for rest and sleep when childrenwere sick. They were more effective in demanding from their husbandsa greatershareof availableresources for their children rather than for their husband's relatives. When sickness did occur, they were more likely to adopt effective home action. This is an important point, for home care is reportedto constituteatleast half of all treatment in theThird World.57
55. Mohamedel Awad Galal el Din, "The Economic Value of Children in Rural Sudan," in The PersistenceofHigh Fertility:PopulationProspectsin the Third World,ed. John C. Caldwell (Canberra: Australian National University, 1977), 2:617-32; idem, "The Rationality of High Fertility in Urban Sudan,"in ibid., 2:633-58. 56. M. E. Khanet al., InequalitiesbetweenMen and Women inNutritionandFamily Welfare Services: An In-DepthEnquiryin an Indian Village,Population and LabourPolicies ProgramWorkingPaperno. 158 LabourOffice, 1987). (Geneva:International 57. N. A. Christakis andA. M. Kleinman,Illness Behavior and Health Transitionin the Developing UniWorld,mimeograph(Cambridge,MA: Harvard versity, School of Public Health, 1989).

LEVELSIN DEVELOPINGCOUNTRIES MORTALITY

59

In south India, we studied in detail the various interrelationsbetween sick childrenandtheirfamilies on the one handand the medical service on the otherin a situation where one village in a ruralareahad a government health center with a resident doctor.The personwho firstdetectedchild sickness was in 80 percentof the cases the mother; however, illiterate mothers were unlikely to take action or even draw attention to the sickness, waiting for their mothers-in-lawor husbands to take note and action. One reason that mortalitywas higher in the south Indian research area thanin the Sri Lankanone was thatonly 10 percentof mothersin the formertook treatment action on their own responsibility, comparedwith 50 percentin the latter.As a mother's education increased, she was more likely to be the chief proponentof action when her children were sick and morelikely to ensurethattheywere treated by the doctor. In the same study, one of the steepest differentialsby maternaleducationwas in the time spent by the motherwith the doctor. Given the absence of backup laboratory testing, diagnosis depends to a very large extent on case histories as presented by mothers. Doctors think that illiterate women cannotadequatelypresentsuch evidence and make relatively little effort to listen to them. Partlybecause of their lack of educationandpartlybecausedoctorssay less to them,illiteratewomen areless likely to carryout the doctor's instructions properly and less likely to persistwith the treatment.Avery steep andsignificantdifferential by educationis foundwith regardto the mother's reactionwhen the child's condition does not improve.With more schooling a motheris increasinglylikely to return

to the healthcenterto reporttheproblemto the doctor, while an uneducated mother frequently fails to do so partly on the grounds that the doctor has already done his best and partlyon the groundsthat she cannottell an important manhe has failed. When we contrastedthis situationwith Sri Lanka,with its much higher levels of female education,we foundin the Sri Lankan household an almost competitiveattitude to the quick detectionof sickness and the seeking of treatment. The strongest contrast,however, was in the Sri Lankan thatwas not reimpatiencewith treatment in and sulting improvement the consequent changing of doctors or from doctors to hospitals after only a few days. The low mortalitylevels suggest that this is an effective treatment strategy.
THE DIRECTIONOF CHANGE

In most of the Third World, with its limited and only slowly spreadingmodem healthservices,mortalitylevels can be dramatically reducedby behavioralchanges. Those changes are not easily achieved, as they affectnot only mortalitylevels butthe structure of society and all social relations. Nevertheless,thereis a potentialfor rapid change that did not exist in the West because Westernsocial patternsare spread, largelywithoutthataim in mind,by education, the media, and religious proselytizing. Educationhas hada majorimpact,and this is now being supplemented by the women's movement. Underlyingit all is the of themarket development economyandacmovements companying awayfromthefamily control patternsand resourcepriorities characteristic of subsistence agriculture.