P. 1
Model Life Tables for China

Model Life Tables for China

|Views: 27|Likes:
Published by Aja Nasrun

More info:

Published by: Aja Nasrun on Feb 26, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

02/26/2012

pdf

text

original

MODEL LIFE TABLES FOR CHINA Jiang Zhenghua

Paper number 0035

Population Research Institute Xi'an liaotong University Xi' an, Shaanxi Province People's Republic of China

1990

Preface

Model life tables provide an important tool for studying demographic forecasting population development as patterns of mortality change.

problems and demo-

In international

graphic studies. mortality in the population of China has received very little attention. Neither the model life tables developed by the United Nations in the 1950s nor the regional model life tables developed by Professor A.J. Coale have been applied to the life tables of China. Fragmentary mortality data from the populations of Taiwan and Hong Kong are

nowhere near enough to represent most of the People's Republic of China. Although the model life tables for developing countries constructed by the United Nations in recent years have incorporated mortality data from the Far East areas. none of these life tables are from

the mainland of China. It is important to include China, with more than one fifth of the World's population, in such analyses.

Before the 1950s China did not collect demographic monality data in a nationwide coordinated manner. Since the founding of the People's Republic of China, some age-

specific mortality data have been available through registration of vital data. In general, most of these data are believable, although the number of deaths at some ages, especially at the youngest ages, are obviously under-registered representative for all sub-regions. In addition. the data were not fully

From 1973 to 1975, a large scale survey of deaths due the causes of specific of

to cancer was conducted by the Public Health Ministry to investigate

mortality rates for the whole of China. Statistical data which were quite representative the general mortality characters for each region were collected for the first time. These survey data had deficiencies, however, and cenain reservations

were expressed by medical in comparison

and public health statisticians on the reliability of the results. Nevertheless, with the previous situation, these data obviously constituted some progress.

The population census conducted in 1982 provided valuable statistical material concerning mortality. This was the first time in China that information on deaths in the
- 1-

whole country was collected. Specific appendices were used to record detailed personal materials relating to the deaths. After the census, a sample survey was conducted to test the quality of the data and proved that the reliability of the mortality data was very high. This provided a solid basis for studying demographic changes and socio-economic development in China and also for filling in a void in information available to international scholars of demography. The China Classified (Regional) Model Life Table Compilation Committee was established in 1984. Mr. Chengrui Li, who was the director of the

administrative office of the third population census under the State Council and head of the State Statistical Bureau, was the chairman of the compilation committee, which consisted of experts and personnel in relevant departments. First of all, according to a unified procedure controlled by the third population census office under the State Council of China, 3136 original life tables (half for males and half for females) were established by local population census offices and electronic computer centers of various provinces, autonomous regions and municipalities. Using the material from the census, sample survey and vital registrations. Professor Zbenghua Jiang, director of the Population Research Institute of Xi' an Jiaotong University, and Mr. Jiyu Luo, associate chief engineer of the electronic computer center of the Sichuan Provincial Planning Economic Commission were in charge of the compilation technique for the model life tables. The Population Research Institute of Xi'an Jiaotong University and the computer center of Sichuan Provincial Planning Economic Commission cooperated in the detailed calculations. In the process of compilation. the original life tables for each region of China were given a quality assessment and careful screening. 1926 tables (half for males and half for females) passed the evaluation and became the basis for the classified (regional) model life tables. The basic data in these tables came mosdy from China' s population census in 1982. A few of the tables came from earlier survey data. such as vital registration data and the cancer survey from 1973 to 1975. By mathematical processing and analysis of a large part of the original
data, the patterns of mortality for the Chinese population could be classified into five
2-

An appendix regarding the use of the model life tables is attached for the users' convenience. II. and also provides a reliable demographic basis for further research work in various fields. the compilation committee of the model life tables invited experts from relevant units to make a formal appraisal of China classified (regional) model life tables in Beijing. South-west China. etc. to identify the mortality pattern for that area and to classify it into a certain family. as well as nationality and other conditions.families. the Medical Research Institutes. improvements were made according to these suggestions. The preliminary version of the model life tables was developed in 1987 and the compilation committee invited experts from relevant departments to evaluate the output in Beijing on March 17 of that year. the 710 Institute of Aviation Ministry.e. categories) are listed in this book. the 710 Institute of Aeronautical and Space Ministry and the . the Beijing Medical University. the software may be used. On December 26. Given the pattern of mortality and one of the parameters of the life table for a certain area (such as life expectancy at certain age. as well. of which 180 tables each are for males and females with life expectancies ranked from age 40 to 75. Software applications have been developed. After the meeting. infant monality rate. Expens and professors from the Technical Economy Social Development Center under the State Council. the Applied Mathematics Institute and the System Institute of the Chinese Acad- emy of Sciences.). 360 model life tables in five "families" (i. From a life table for a certain area. and V. and other related agencies gave an affinnative evaluation on the preliminary version of the China classified (regional) model life tables and made several suggestions. ill. 1987. Xinjiang autonomous region.3- . North China. This classification is quite consistent with regional social and economic characteristics. the State Statistical Bureau. Middle and East China. The development and compilation of the China classified (regional) model life tables has taken four years. North-east China. a corresponding life table may be produced. The five families include: 1. Experts and professors from the Applied Mathematics Institute of the Chinese Academy of Sciences. IV.

comments and suggestions. be somewhat inadequate to reflect some prosperties of the mOrtality of the Chinese population. electronic computer centers (stations) and vital registration agencies of various provinces. The model life tables compilation committee and development working group wish to convey to these officials our sincere thanks. As more mortality data become available. the technique applied in the research is rigorously scientific.State Economic Infonnation Center and other agencies. there has been rapid change in the socio-economic conditions of our country. and is original compared to methods used by foreign countries. Chengrui Director of China classified (regional) model life table compilation committee August 15. The appraisal cenificate given by the appraisal committee is included in the first part of the book. China classified (regional) model life tables may. the Public Health Ministry and Tumor Research Institute. autonomous regions and municipalities. we enjoyed energetic support and assistance from population census offices. we intend to fwther revise and improve these model life tables. In the short history of collection of mortality data in China. Li. In the process of developing and compiling of the China classified (regional) model life tables.4- . therefore. Seven experts and professors were invited to act as an appraisal committee headed by Professor Shouren Wang. 12 units altogether. Users are welcome to offer criticisms. etc. Members of the appraisal committee and relevant experts directed questions which were answered by the working group The appraisal committee then met and reached a unanimous conclusion that the reliability of the original data selected for establishing the China classified (regional) model life tables is quite high. 1988 . participated in the meeting.

Sun Yiren. Wang Weizhi. Dong Qing. Ma An. . Chengrui Vice Chairmen. Luo Jiyu Members: Li Nan. Sun Jingxin. Xing Ying. Su Xiankang. Qi Xing. Wang Yingluo.The List of Numbers of Compilation Committee of China Classified (Regional) Model Life Tables Chairman: Li. Ding Diji Members: Jiang Zhenghua. Zha Ruichuan.5- . Shan Ming. Luo Changrong.LuiZonghua . Shen Yiming. Mei Xiangfu. Xiao Zhenyu. Yu Wang. Zhou Qifeng. Wang Shouxiang Compilation and Working Group of China Classified (Regional) Model Life Tables Group Leader: Jiang Zhenghua Vice Leaders: Wang Shouxiang. Fan Murong. Li Xiru.

The corresponding regions for these families were given as follows: The first family: South West China. The Appraisal Committee was organized by: The State Statistical Bureau of China.1987. December 26. Date of Appraisal. Briefintroduction to the research on the classified (regional) model life tables of China. A. a. technical and users' manual. related applications software. The raw materials used for developing the classified (regional) model life tables are 3136 sex-specific life tables(l) compiled from the 1982 population census by census offices and computer centers of different provinces. the second family: Middle and East China. The Computer Center of Sichuan Provincial Planning Commission and Economy Commission. data on magnetic disk. autonomous regions and municipalities. b. Raw materials also include the sample survey data collected since 1958 and 50 life -6- . The classified (regional) life tables of China include five families of sex-specific life tables basically classified by regions. The outputs of the research provided to the appraisal committee were: printed tables. The Population Research Institute of Xi' an Jiaotong University. the fourth family: Nonh East China.Certificate of Appraisal for the Classified (Regional) Model Life Tables Research and Established by: The Leading Group Office of the Third Population Census under the State Council of China. the fifth family: Xinjiang autonomous region. the third family: North China.

The concentration of the parameter ~ illustrated the gross \. The method modified by the authors is called the optimized K mean method. where L(1 ) represents the Logit transform of a \. model building and parameter identification. the classification of families of life tables was completed. Taking L(1J as the Logit transform oflife table parameter 1. and J3 in Logit transform represent the level and pattern ofmonality separately. the centers of the families could be formed with each family and center corresponding to specified regional character. The model developed in this research was: A 2* = (Co + CITt + C2T ) a qt t 't . The well-known Legit transform was applied primarily for clustering. then L(1J =ct + J3. The research that produced the model life tables was divided into three stages: cluster analysis. t >I< >I< standard life table parameter 1 . The parameters a. establishment of the model life tables.7- . c. Finally. >I< . The cluster analysis was used to determine the geographical classification of the pattern of mortality in China by identifying the mortality patterns of the origina1life tables and to give the changes by age-specific rates corresponding to the different levels of mortality.tables(2) compiled from data of the nation-wide cancer survey conducted in the period from 1973 to 1975. The average of each group of life tables gives the primary center of the corresponding mortality family.L(1 ). The corresponding technique for the formal problem is cluster and model life tables -to form the centers for different mortality families. In this way. geographical distribution of the original life tables in different families was tested and few original life tables inconsistent with the geographical distribution were detected. classification of the huge amount of the original life tables. According to the principle of minimizing the Euclidean distance of the analyzed life table to the center of the member family and maximizing the Euclidean distance of the life table to centers of other families.

The project was supported and' fmanced by the Office of the Third Population Census Leading Group under the State Council of China. The Population Research Institute of Xi' an Jiaotong University and the Computer Center of the Planning and Economy Commission of Sichuan Province have done the research work jointly. and C2. to serve socio-economic construction of China in wider and deeper Note: (1) Collected by the Office of the Third Population Census under the State Council of China and the Computer Center of the Planning and Economy Commission of Sichuan Province. the parameters corresponding to integer live expectancy at birth of 40 to 75 years. is Co. China is a huge country with great differences in sub-regional geographical.where CJ"* and 1\ ql are the parameters of the standard life table and that of the identified life table. the procedure to calculate parameters simply to solve linear equations. d. None of the five model life tables published previously made use of Chinese data in the research process. population projection and planning in different sub-regions. C}. CI. However. the model life tables could be easily established. e. The research also enabled different population survey data ways. The Research of China Classified (Regional) Model Life Tables was a largescale application of the 1982 China population census data. (2) Compiled by the Cancer Research Institute of the Medical Academy of China. Applying the model mentioned above again. and C2. Moreover. Fitting smoothing curves to identified parameters Co. The research project began in June 1984 and ended in September 1986. The model has an excellent character of equal first three moments for the identified and original life tables. It is essential to establish China classified (regional) model life tables using Chinese data.8- . The model could be' applied in demographical analysis. socioeconomic conditions. .

education and health care. Comparing it to similar work. The large scale of the research was rare in the world. . The Minutes of the Appraisal Committee of "China Classified (Regional) Model Life Tables" Referring to the compiling Committee of China Classified (Regional) Model Life Tables. and reached the following conclusions: The research on the relationship between the levels and patterns of mortality is closely related to work in many fields. 963 original life tables were accepted for each sex. The methods applied in compiling model life tables were strictly scientific. The original data used in this research are of high reliability due to the strict selection procedure. As a result. the Appraisal Committee had tested the research output of the project "China Classified (Regional) Model Life Tables" in Beijing on 26 December 1987 and accepted the summary report of the working group. A lot of machine time was spent and researchers from different departments and academic fields cooperated to analyze the data and compute the outputs. Although research in this aspect has been done in other countries several times. This is the first opportunity to establish the China Classified (Regional) Model Life Tables successfully.9- . the compiled life tables can be usefully applied. The members of the Appraisal Committee and related specialists raised points for argument and discussion. abroad. This leads to a relatively high error in China's population projection and data analysis by applying indirect methods to determine the effects of the change in mottality level on the population age and sex structure and other related socio-economic problems. The census data of one billion population and other historical materials were tested.B. Thus. none of these made use of original Chinese data. The outputs of the research has supplied a reliable basis for the research in the fields of demography. and will attract the attention of population scholars everywhere. the research has a unique character of its own and is of excellent quality. The research filled an important gap in China population studies.

It is also necessary to publish related computer software as soon as possible so that application of the methods can become widespread. The cormnittee suggests that a revision be made when more new data become available in the future so that the results could be applicable to new situations. The conclusion of the State Statistical Bureau of China agrees with the conclusion of "China Classified (Regional) Model Life Tables" made by the Appraisal Committee. The Appraisal Committee of "China Classified (Regional) Model Life Tables" 26 December 1987 c. 20 December 1987 - 10 - .The research mainly used the 1982 population census data.

The List and Signatures of the Members of the Appraisal Committee to "China Classified (Regional) Model Life Tables" Name Chairman Wang Shouren Working Unit Applied Mathematics Research Institute of Science 710 Institute of the Aviation Ministry Beijing Medical UDiversity Population Research Institute of Beijing Economy School Population Statistical Department of the State Statistical Bureau Economy Research Institute of the State Planning Commission State Economy Information Center Title. Vice Chairman Members Yu t Jingyuan Li TIan1in Xu . LiHuijing . Zhang Dayang .11 - . . Shaoyu ShenQiuhua . Post Research Fellow (Professor) Research Fellow (Professor) Professor Associate Research Fellow Deputy Head and Associate Professor Associate Research Fellow Senior Engineer .D. .

Qiuhua LiHuijing . . Lecturer . .. Post Fonnal Director and Professor Director and Professor Director Research Fellow Research Fellow Professor Associate Research Fellow Associate Research Fellow . Ruichuan Shen. . Associate Research Fellow Professor Deputy Director and Associate Professor Associate Research Fellow Associate Research Fellow Deputy Director of a Division Zha . Xu . Research Institute of The Social Science Academy of China Population Research Institute of the Chinese People' s University Population Statistical Department of the State Statistical Bureau Economy Institute of the State Planning Commission Economy Information Center of Sichuan Province State Economy Information Center Population Research Institute of Beijing University . China Population Infonnation Center Senior Statistician Population.12 - . .Shaoyu UTlanlin Wang Sulin Xiao Zhengyu WangWeizhi . Shen Yiming Wang Shouren YuJingyuan . XingYing SunDaqi Wang Guoyi ) .E. The List of Participants of the Appraisal Meeting of "China Classified (Regional) Model Life Tables" Name Li Chengrui Jiang Zhenghua J Working Unit State Statistical Bureau Population Research Institute of Xi' an Jiaotong University Population Statistical Department of the State Statistical Bureau Applied Mathematics Research Institute 710 Institute of the Aviation Ministry Beijing Medical University Population Research Institute of Beijing Economy School State Economy Information Center Title. .

D.Name Yao Keqin Wang . . . Population Statistical Department of the State Statistical Bureau China Population Information Center State Economy Information Center . .13 - . Wang Shouxiang UBohua YaoFeng Working Unit Planning and Statistical Departtnent of the Health Ministry of China Beijing Medical University Population Research Institute of Xi'an Jiaotong University Title. Renan LiNan . Post Chief Doctor Associate Professor Ph. Student Deputy Director of a Division .

cultural background. They are a set of uncrossed curves shifting gradually from a J shape to a U shape. If a family of mortality . and correspond to a lower standard of social development process. In this case.14 - . China is a populous country with many races. Along with increasing social development. Cluster Analysis on the Original Life Tables I. The mortality pattern and level could also be presented in terms of survival probability. Different populations with great differences in socio-economic and demographic conditions may differ a lot in age-specific death probabilities. the mortalities of the populations are said to be at the same level but to have different patterns. Figure 1 gives the curves of age-specific death probability rates corresponding to different stages of development in a population. A family of survival probability curves corresponding to the death probability curves of figure 1 is illustrated in figure 2. health and medical care. Obviously. The curves far above the horizontal axis represent a lower life expectancy and higher mortality. probability curves given in figure 1 has similar shapes and do not intersect one another. The Principles for Cluster Analysis on the Original Life Tables It is generally accepted that the level and pattern of mortality is basically determined by the biological condition of the population living in a certain area. we say that the family of curves has the same pattern but different levels of mortality. environment.The Technical Description of the Establishment of the Classified (Regional) Model Life Tables for China I. There are big differences in culture and social customs among different races and regions. etc. However. agespecific death probabilities for all ages in an area will decline. the curves which represent higher life expectancy and lower mortality are closer to the horizontal axis and correspond to a higher level of social development. The mortality conditions may also differ from region to region due to the differences in geographical environment and economic . mortality is also strongly affected by external factors like socio-economic condition. even though they have the same life expectancies at birth.

.• 1n)' the following relationship can be used: Logit (Ii) = a + ~. The purpose of doing cluster analysis on the original life tables was to identify several different families of mortality pattern from the mixed original tables. 2.. The results of the cluster analysis could be applied to decision analysis and research in socio-economic and other fields. Brass.. 13. Logit (1... 13... The Cluster Analysis Method Both the survival probability rates and age-specific death probability rates could be used in cluster analysis of the original life tables.. In this case. ... . -x Here a and ~ are constants which can be found by the following procedure: .development. (1) where Logit is a function defmed by Logit (x) = 0.5 In IX . 2. . if two life tables are expressed by age-specific survival probability rates (l ]. To populations living in regions far apart.. In) and 1. 3. The Logit transfonnation was applied for preliminary classification. these small groups will be classified as belonging to other families. The classification procedure was divided into two stages and tested by graphical and computational methods.. 12. Small populations within a larger area may have a pattern of mortality which differs from that of the majority of the surrounding population. the factors that determine mortality are more or less the same. Each family of life tables represented a pattern of mortality typical to a certain region of China.15 - .. (l.. so that the differences in the pattern of mortality may also be great This suggests that it should be possible to divide China's population into several groups according to the pattern of mortality in geographical areas. the factors determining mortality may differ a lot.1 ) i-I.. so that the patterns of mortality in these regions should be about the same. .. To populations living close together.• n . According to W.

According to formula (1)..Logit (1 ) '" n By the minimum least equal error method. B= i=l L [Legit n n (li)] n . 3•. [Logit (1. 2.I.. logit .. - 16 - .A]2 1 and where n .. we have n L ~=--------------------n * i=1 i=I [Logit (1.• n). 1 i=1 A=----- L n . tan 9 = ~ and a is the intersection of the line on the vertical axis (see Figure 3).A] [Logit (Ij) . ~ is the slope of the line found by one-dimensional linear regression. ) . Given n points [logit (I. we have Logit (11) = a + ~...B] 1 * L [Logit (1. ) . (Ii> ] (i . ).e. i.Logit (1*) 2 • • Legit (In) = ex + ~.Logit Logit (11) '" (1v = a+ ~.

Age x Figure 1 Age specific death probability curves 1 1 x o L_ ~~~~-Figure 2 Age specific survival ratio curves .17 - Age x .

as and ~s can be identified with those of a standard life table. The differences in the pattern of mortality among these life tables can be recognized by comparing the values of~.: •/ . the classification mentioned above is only a preliminary one.... a reflects the difference in mortality ~ of two life tables and ~ the difference in pattern of the life tables. set of life tables with the same level of life expectancy. .j" .. Q• . ". Similar patterns of mortality for two life tables will result in ~s that are close together.18 - . '~': .. the parameter ~ can be used as a crude classification on the pattern of mortality for a set of life tables with the same level of mortality.:/ )/ . Graphical illustration of the parameter identification In this calculation.. then we will have a = 0 and ~ ::::: Taking the legit transform of survival probabilities of a 1.." ./ . if two life tables are exactly the same. o Figure 3. A life table with higher infant mortality and lower old age mortality will produce a smaller ~ than a life table with lower infant mortality and higher old age mortality. Therefore. More accurate classification was done by applying the K-mean method with the information from the preliminary classification.. In particular. The traditional K-mean method was modified . Since only two parameters are used to express what is in reality a non-linear mortality curve.

This then defines a stable K-family of original life tables.. bn) .. they form K demographic groups. the grouping of the vectors is then adjusted by classifying them within the class with minimum distance between the center and observed vectors. Comparing the distances from each vector to different center vectors. This modified K-mean method was named the Qprimized K-mean method. Define the distance between the two vectors as It is obvious that the smaller this distance between two vectors is. These vectors should be the centers of different groups.. The new classification could serve as a reformed original class in the next classification procedure.19 - . In mathematical language.. The optimized K-mean method was applied to find several curves in the space which could collect all the points around them with the minimum distance. the original life tables are points disttibuted in an ndimension space. These curves form typical model life tables of several families. If a set of a-dimensional vectors can be clustered into K classes. dividing the sum of elements with the same level of mortality in a group by the number of the vectors in the corresponding group. .. a2•.by the authors to minimize the uncertainty of the classification.• an) and B = (b]. the closer the two points A and B are. The procedure mentioned above is repeated until each vector is placed in the class with minimum distance between that vector and its own center. new vectors could be formed for each group of original life tables. A stable K families of original life tables were considered to be satisfied. b2. Let two vectors of n dimensions be A = (aj.. Based on the preliminary classification.

i. Considering age-specific survival probabilities of each original life table as a vector. It was found that among life tables with life expectancy of 65 to 69 years for males and 68 to 72 years for females. All the original life tables could be represented by the standard life table with two parameters of a and B.. sum up the corresponding elements of all original life tables in a segment and divide by the number of life tables in this segment. As expected. Each segment contains original life tables coming from almost all provinces and autonomous regions. All of the different patterns of mortality in China could be identified from each segment. It is quite possible that life tables with different patterns but the same levels may be clustered in a group according to level rather than according to pattern. the original life tables in each segment have the same rounded value of life expectancy at binh. autonomous regions and municipalities except Xinjiang and Tibet. Data Processing and Cluster Analysis After reliability testing. the space of life tables was divided into segments by integral digits of life expectancies.4 years for males. the life expectancy at birth was 66. the lowest and highest life expectancy at birth are 43 and 73 years for males and 43 and 78 years for females. the original data from this region were taken as a separate family to deal with. the original life tables from the third census data were more concentrated in five segments of 6S years to 69 years for males and 68 years to 72 years for females. Since the mortality pattern of Xinjiang population was very different from the others. 1936 original life tables for males and females were accepted. The rest of the life tables represented the mortality patterns of all other provinces. Thus cluster analysis could be done among life tables with the same level of mortality. preliminary classification on the pattern of mortality of the original life tables could be done by identifying the values of ~ of different life tables with a standard life table. Applying the Logit transform method mentioned above. An abridged life table could be established in this way as a standard. four groups could be identified in every . The details follow. and 69.20 - . To remove this problem. According to the population census of the year 1982. Among these life tables.3..e.4 years for females in the year 1981.

North and North-West China. The . to produce the final classification of original life tables. It was necessary to remove the life tables with clearly different characters from its own family by direct observation. graphical comparison and minimizing the distance between the life tables to the center of the family. whole families could not be fully covered. The four regions are: South-West China. The above procedure gave a preliminary classification and determined the number of families as K=4.segment according to the values of~. This means that the mortality patterns of China could be classified into four families except for Xinjiang.21 - . Middle and East China. Because few of the original life tables had very low or very high life expectancies. The preliminary classification is based on the values of ~s only. after adjusting and testing the validity of regional classification. the center vector of each family of life tables serves as a model of the mortality pattern of the family. the characteristics of the original life tables had to be tested before they could be classified into a certain family. The vast majority of the original life tables belong to four separate regions. The geographical classification for the mortality patterns of males and females was perfectly consistent. Different groups of life tables correspond to different regions. The socio-economic and geographical conditions of different sub-regions in a province or autonomous region may differ so that life tables may differ in many ways even among those with the same life expectancy and some of these differences may be large. This procedure was applied to both males and females life tables. All the removed life tables were treated separately. Thus. The optimizing K-mean method was applied to the preliminary classification results. and the North-East China. it is not accurate to describe the characteristic difference of whole life tables with one parameter. After identifying the classification of mortality pattern and corresponding regions. The rest of the life tables of each family in the segment were classified again by K-mean method. Obviously. The great majority of the original life tables in a group came from the same region in China.

Thus. Xinjiang autonomous region has a special pattern of mortality. and 154 original life tables for the fourth family of model life tables. 358 original life tables were used to create the first family of model life tables.22 - . 207 for the third family. A special family of model life tables was established separately on the basis of original life tables in this region. five families of model life tables were found for the Chinese population by cluster analysis. the original life tables which were accepted in order to create the regional model life tables of China included 963 for males and females separately. Ranking the life tables by life expectancy. four families of model life tables were found Each family of model life tables represented a typical mortality pattern in China After removing life tables with large variations.procedure of removing and classifying was repeated until all the life tables in each family were consistent with the geographical locality. . 244 for the second family. the center vectors of the same region in different segments formed a family. Results After identifying the center vectors of different families in each segment. Among them. 4.

al' . of the observed life tables whose differences in life expectancies is less than one year. Tm be m preliminary standard life tables.. Let tl. "'.n. the weighting coefficient corresponding to age x. t2.1) This means that the probability of dying of an observed life table can be expressed as a function of that of a standard life table. '" m) is the arithmetical average ...2. Here nqr) is the probability of dying between age x and x+n of standard life table Tk• ll(}.23 - . an are parameters to be estimated. The 1. The geographical areas corresponding The first pattern: The second pattern: The third pattern: The fourth pattern: The fifth pattern: to each category are given below: South-west China Middle and east China North China North-east China Xinjiang autonomous region Each of the five patterns contains a set of model life tables for life expectancy at birth between 40 years and 75 years with one-year increments. By cluster analysis.. Choose a representative table.. each Ti (i = 1. and TI• T2. tm be m observed life tables with the same pattern. the mathematical life table with good inheritence Tk (1 S k S m) as the standard life model of monality pattern we construct is nq~) n<lx = (ao + altx + "'.. the mortality patterns in China were classified into five families according to the resemblance of mortality patterns across geographical areas. Establishment Mathematical Model and Generation of Model Life Tables of the model and identification of parameters: We develop here a more accurate and easily calculated mortality model and apply the model to construct the classified (regional) model life tables of China.. + ant~) (2. tx is .

~ . = 1. aI' .2) Take the partial differential with respect to a. Obviously.. r min xL (nq(t).The character of the model will be illustrated in the section on estimation of parameters. m) according The values of a(t). So the first step is to identify the parameters ai (i = 0. If the variation trend of parameter ai (i = 0...~ t (i+j-I)-N • q(k).• N) is known.(t) in objective function (2. x nx • i=O-N..an are given... a (t»T (N + I) x 1 o 1 N . x q)2 / ( q(k) • t N) nx nx x (2.a(t) with respect to Tl can be obtained by solving the o 1 n following equation.. j=I-N+l b. 1 Note: g ij .N• q(l) 1 xx nx G = (gij)(N+I) x (N + 1) B = (bi)(N + I) x 1 -a A .• n) of each Ti (i to equation (2.1). . . a model table with arbitrary life expectancy eo can easily be obtained. a(t). when a certain series of parameters table with specified life expectancy no.24 - ..2).1). . a model eo can be uniquely obtained from equation (2.. = L ti-I. Then a family of model life tables can be generated.a (t)...«t) .

choose N=2 as our model of generating model life tables of China in five year age group. + aCt) tN) N x nx q(k) is the estimated value of a~l). the parameters aCt) can be derived from this set oflinear equations. In addition.. B 1 Obviously..1). (l)h.. 2. this model is simply to calculate. Increasing the value of txo could make the estimation of n~'L) more accurate. N L ( q(t~ (l)h= L (nq x Ox tx' X x . I. The explanation mentioned above is only a brief description of the general characters of this model. n '" It means that the model has equal Nth order moments of n~t) and o'lx' Now we discuss the selection of standard life table Tk and the weighted coefficient tx_ (1) For reducing error in parameter identification. (2) As the weighting coefficients. we can direct! y get: i = 0.Then A=G-l. Therefore.. a reasonable value of tx can make the estimation errors quite uniform for all ages. As an example. in equation (2. (xl Where. the standard life table T k must represent the general pattern of mortality in its category.. Thus.. n__ (a(t) = n-ix a + a(t)t 1x + . it is a three-parameter model: - 25 - .

0000 0.1746 0.1891 0.2139 0.1 and Table 2.5911 0.5698 -0.2838 0. .26 - .2528 -0.2152 0. I J-L model The preliminary standard life Table Ti The maximum opposite erroa E CoD model I J-L model Tl T3 T4 T5 T2 -0..2189 0.3514 0. where C-D method denotes the Coale and Demeny model and J-L model (Jiang-Li) is the one presented in this book.1 Fitting Errors of Fitting for Pattern 4.2135 T9 1'7 T8 T6 0..1799 . Table 2. males e = max (qj _ ~(l))/~(l) The preliminary standard life Table Ti The maximum opposite CoD model s:rm[St.0976 0.2 present errors of fitting two models to the observed data.1879 0. Table 2.• 17) The parameters can be obtained by solving the following problem: Regarding the accuracy of our model.3953 0. the following example gives a comparison between the Coale and Demeny model and this one.3441 -0.4707 0.Where to = tl = 30 t2 = t3 = 10 tk =1 (k = 4.

Tables 2.0621 0. and 0. Among the age-specific death probabilities.3441 0.0854 ~.1456 0.1 and is the maximum of the relative errors in fitting age-specific death probabilities to the original life tables.1891 0.3342 0. our model is a little inferior to the Coale-Demeny model. the fitness of infant mortality rate by using our new 27 .2139 0.284 by using the Coale-Demeny model for the third original life table of the fourth family for males.0837 by using the Coale-Demeny model and 0. For instance.0434 -0.2 .175 by using our new model. the maximum relative fitting errors of two tables is inferior by using our model than by using Coale-Demeny. of the fourth family of males was the worst among the five families. the infant mortality rate is very important.1952 0. the relative fitting error of the infant mortality rate is 0.2 show that among nine original life tables of the fourth family for males.0544 Among the center life tables which were available at the beginning.2989 0. The results for the comparison between the worst fourth family of our model for males and the Coale-Demeny model was taken as an example The maximum relative error is given in table 2.1144 0.1 and 2.0837 -0. Only for one table.2056 0.Fitting Errors of qo for Pattern 4. the fitness.0514 by using our new model. males The preliminary standard life table Ti T1 Eo CoD model I J-L model The preliminary standard life table Ti T6 1'7 T8 T9 Eo CoD Model I J-L Model 1'2 13 T4 T5 ~. the maximum relative error is 0. and for another table.0514 0. Taking the third original life table as an example.0000 0.2528 0.2. and higher identification accuracy is required in estimating IMR. Thus the absolute value of relative error between infant mortality rates for the original fife table and the estimated life table should be smaller. The relative errors of fitting the life tables in the fourth family are given in table 2.Table 2.0002 0. Various types of data were used to compare the identification accuracy of our model with that of other models.

high accuracy and easy calculation. and a2 for the first family model which was established on a large amount of original data. and ao. In light of the characters of parameters ao. This new mathematical model has the characteristics meaning. special methods are needed to determine the trace of the parameter curve for life tables with high and low life expectancy. and a2 reflect the characters of different family center life tables. and a2 + a2tx2) s~k). So that we may conclude that our of clear model has quite high accuracy. and a2 will lead to a lower life expectancy of the corresponding life table.a standard table T k could be chosen to identify the parameters of m group.. a}. we have very limited numbers of life tables with very high and low life expectancies.• T m. T 2•. From equation sqx = (ao + altx ao. The Establishment of the Model Life Tables Using original center life tables T 1. al. once the standard table Tk and weighting coefficient tx were determined. al. We call this the property of monotonicity. really the process of searching parameters The process of establishing life tables is a2' The changes of ao. In particular. could be found. the death probabilities were fully determined by and a2' Higher values of ao. ao.3) - 28 - . follows: k Y = 1 + ea+bx + The logist curve is described as e· K>O (2.. the logistic curve was found to be good for fitting the observed relationship between parameter and life expectancy.model is inferior to that by using the Coale-Demeny model. a 1. a 1. The effect of changing parameters could be found by fitting a curve to the parameters. The trend of the changes in three parameters. Therefore. But the identified parameters do not necessarily have a monotonic relationship with the change of life expectancy due to the limitation of the number of center life tables and the consistency of the patterns. 2. al.

m) .6) .4) written as k ---1 y-a =ea+bx let a = min (a~) . (2) b <0. we have min F (a. It will be monotonically increasing when b > 0 and monotonically decreasing when By using this curve.f (aai»2 (2.5) aO - 9 m so that for the m original center tables.we obtain -1 + _k_ = ea+bx.51 K = max {a~) } .3). (2. + ~ at.29 - . K>ao-9 Take logarithms of (2.4) a + ben = In ( k -1 ) = f (ac). y-9 If we fit the relation between ao and life expectancy at birth. By changing the form of equation (2.It has the following 2 characteristics: (1) The upper limit is k + at the lower limit is a.min {a~) } . Obviously: 3z > 0 are adjustable parameters.8. but also the determination and the adjustment of the parameters for model life tables with high and low life expectancies at birth can be made in great convenience.b) = min i=l I (aa + beoi . not only the monotonic change of the parameter with the life expectancy at birth can be assured. ke l . then the equation should be (2.

30 - . Similarly. is calculated through regression extrapolation. child and youth mortality may need to be adjusted to keep the consistent and even change of death probabilities in one family of model. coefficient is as follows: . For developing model life tables with low life expectancy at birth. among which the ratio q()"qi> for the life table of e~ = 40.7) . 14of the model tables with e~ under that of the standard table must be adjusted according to the principle of even change. This form is decided by the characteristics of the clusters of original life to. The weighting coefficients of the standard table and tables with e~ greater than that of the standard table are kept as constants. the relations among infant. This may be realized by adjusting weighting coefficients ao and e~. t10t2..b) for a and b and set them equal to zero to obtain: (~)= Leo N m i=l m :Leoi m Ie2 i=1 i=l -1 m i=l L !(aoi) (2. 01 m '=1 L !(aoi) eoi The determination of parameters a and b will lead to a cenain fonn of relationship between tables. and 14appropriately.. t3. .. the relation curves of parameters al and a2 to e~ can be fitted. The adjustment formula for the weighted. Not ony the appropriate adjustment of the model for e~ = 40 is needed but also the weighted coefficients to .Take partial differentials of F (a.

Attachment: Symbol explanation of the model life tables Qx Ix Dx .i = 0.• 4 (2.cumulative number of person years of age x and above.number of deaths in age group x.. a1136 mofel life tables with integral digit life expectancies e~ from 40 to 75 years and life tables can be developed.the death probability of age x.number of person years in age group x. . . The range of life expectancies can be extended if necessary. ti is the unadjusted weighting coefficient.number of persons alive among 10.8) where ~ is the adjusted weighting coefficient.000 live binhs at exact age x. Lx Tx Ex . t~40) is the weighting coefficient for e~ = 40.. In this way. . . Any life table with non-integral digit life expectancy from 40 to 75 years could also be generated as needed.31 - . . e~) and e~ are the life expectancies of the standard life table and rnodellife tables to be generated respectively. -life expectancy at age x.

74. Example: Assume that the monality pattern of some region confonns to pattern I for females in model life tables of China. projection and other studies. Step 1. x+S) as 5q~5and 5q~. years. . the corresponding life table can be calculated by using interpolation. respectively. They are used in a variety of ways and the details can best be illustrated in concrete applications. I.32 - . Denote the probabilities of dying during (x. Find two model tables in pattern I for females whose life expectancies at birth are e~5 = 65 and e~ = 66 years respectively. The Application of the China Classified (Regional) Model Life Tables Since they combine levels and patterns of mortality.Appendix I. Let eo. and the life expectancy at birth. and are widely used in population analysis. Calculation of a corresponding life table with desired life expectancy at birth If the pattern of mortality and life expectancy at birth are known.74 e: = 65. model life tables simplify many aspects of population studies. some of which we now present All calculations illustrated below are for quick estimation by hand. is 65. More accurate calculation can be done using the application software developed by the authors. Calculate the corresponding life table. as shown in Tables 1 and 2.

81 6.06253 .05975 .04225 .25 13.51 55.72 42.01237 .16206 .00943 . 97240 6500000 377133 .15521 .24 46.04056 .01602 .10657 . .3 . 00408 .96 24.01 60.56 65.03332 .00777 .01669 .72 47.00000 lx 100000 96251 93337 9']1574 92295 91578 90715 89739 68629 67209 65551 .44808 1.19 4.01981 .01901 .72 16.00 67.22209 .2914 663 379 717 863 976 1110 1420 1656 2338 10464 12650 14264 13133 16911 3375 4770 7646 97488 379174 465025 462422 459685 455733 451135 445921 439596 431901 421912 407630 387267 356227 310950 253164 185678 117387 71504 6600000 6502512 6123338 5658312 5195890 4736205 4280472 3829337 3383416 2943819 2511918 2090007 1682377 1295110 938883 627933 374769 188691 71504 66.34 28.01120 .76 29.93 13.03026 .74 33.06 56.02847 .19 8.Table 1 x 0 15- Qx .12 21. 63213 79638 75066 67422 56958 44306 30044 16911· Ox Lx 'I'x E'x 510152025- 455055606570758085+ 40- 3035- 3749 .29 4.00449 .10186.02733 .07 17.01075 .0.78 65.32194 .00 66.~2 8.43711 1.36 25.33 - .30 51.03749 .00865 .46 6.01289 .3l Table 2 x 0 1- Qx .67 38.65 10.00000 lx 100000 95881 92686 91952 91539 90748 .93 11. 6402760 461594 6025627 458728 5564033 455717 5105305 451509 4649588 446764 4198079 441386 3751315 434864 3309929 426935 2875065 2448131 416649 401985 2031482 380987 1629524 349046 1248536 302710' 899491 244136 596781 ·352644 177007 110001 175637 65636 65636 65.77 51.33211 .74 2.19 42.22 37. 89856 88850 87705 86241 84533 82127 78657 73738 65880 55204 42451 26352 15648 Ox Lx 'IX E'x 10152025- 30354045505560- 6570758085+ 4119 3195 734 413 791 892 1006 1145 1464 1708 2406 3470 4919 7858 10676 12753 14099 12704 15648 .23102 .17 33.00710 '.00983 .60 61.00791 ..04119 .2. .

01922 .. as in Table 3.01619 .26 25.01250 .16 51. an appropriate pattern of model life tables can be selected according to socio-economic and geographic characteristics to estimate detailed information about the level and age pattern of mortality in the past.45 60.01067 .06046 .00799 .36 65.00419 .eo ) = -0.85 10.15697 .28.7. Step 2.22 software developed More accurate estimation can be made by using computer by the authors. Example: The age-specific probabilities of dying of a male population in 1984 are shown in Table 4 and the corresponding life expectancy. 66 65 = (eo .15 years. II.eo) / (eo .65 29. Calculate sqx using the following formula s<lx where = a·sqx + (l-u) • sqx a 66 . Estimation of historical data on death In situations where previous information on death is lacking or deficient.55 38.34 - .03845 .. Table 3 x Ox Ix Ox Lx 'IX Ex 0 15- 2530- 101520- 3540455055606570758085+ .00000 100000 96115 93168 92487 92100 91364 90493 89509 88390 86959 65288 62932 79532 74723 67022 56501 43823 29599 16578 3845 2967 661 387 736 871 984 1119 1431 1671 2356 4809 7701 10521 12678 14224 13021 16578 3400 97421+ 376646 464139 461466 456660 454641 450005 444749 438373 430616 420550 406161 385640 354363 306807 250809 6574000 6476576 6097928 5633769 5172321 4713661 4259019 3809014 3364266 183555 115443 69950 2925893 2A95277 20747Z7 1668566 1262926 926563 619756 368948 185393 69950 65.06 33. the desired life table can be calculated according to the series of sqx. If Co at some past year was 65. Then.97 8.22439 .91 56.98 17.04099 .26 4.03106 . estimate the corresponding age-specific probabilities of dying. is 72.10307 .59 47.42 6. eo...02 20. .00953 .32456 .74 67..00731 .43993 1.17 13.02762 .06 42.

008174 0. Select the representative pattern from the model life tables of China.Table 4. Compare the mortality pattern in the given population with each of those in the model tables for males of China.039216 0.15 years in the five patterns of model tables.113725 0.010478 0.001426 0.011824 0.021314 60657075- 55- 80-84 0.295817 0. .068482 0. The age-specific probabilities of dying for these are shown in Table 5.002939 0.002518 0.005678 0.004042 0. x nqx " x 5qx x 5Qx 101520- 5- 0 1- 0.002640 0.180406 0.003517 0. This can be done as follows: (a) Find model tables whose life expectancies at birth are all 72.35 - . and determine which one of the five patterns provides the best representation of the mortality pattern of the given population.002836 25"': 3035404550- 0.436340 Step 1.

07489 0. between probabilities of dying (5<Ix) shown in each column of Table 5 and those in Table 4.09741 0.00605 0.42420 Pattern V males 0. D= (4) q(S) Sq x .00371 0.00465 0.03060 0.00963 0.01828 0. .5.13372 0.02322 0.01476 0.01213 0.00785 0. The results are respectively 1.00437 0. are w hat we need.00593 0.01335 0.14697 0.70 years by using interpolation in the second family of male model tables.00401 0.01879 0.01822 0.00727 0.00501 0.00582 0.00336 0.05665 0. the second pattern.00146 0.6789.5512.04647 0.9807.01906 0. Step 2.06850 0.00269 0.00739 0.00669 0.30550 0. D.10919 0.13572 0.12376 0.00193 0.S x I (--(-4)sqx )2.00401 0.07598 0.13887 0.00206 0.01144 0.09211 0.04525 0.01047 0.01106 0.4794.27149 0.00713 0.00478 0.00304 0.33166 0.02746 0.00610 0. should be chosen as the mortality pattern of the given population.00739 0.00272 0.16456 0.00988 0.00195 0.01300 0. Its age-specific probabilities of dying.00482 0.0.00622 0. \ pattern age \ group\ 0 1510IS20pattern I males 0.01546 0.51165 pattern IV males 0.00258 0.23956 25- 303540455560- 50- 65707580-84 (b) Calculate the relative variation.19215 0.20046 0.29326 0. which corresponds to the smallest relative variation.00807 0.01066 0.00308 0.42093 pattern III males 0.03400 0.21939 0. Find the model table with the life expectancy eo = 65.05295 0.36 - .00391 0.00258 0.00470 0.9161.00528 0.20546 0.03083 0. Obviously.02003 0.38147 pattern II males 0.03396 0.06336 0.Table 5.03842 0.00406 0.0. shown in Table 6.00374 0.0.00442 0.

28557 0.02873 0.08811 0.00630 0.00615 3035404550- 0.00351 0.04809 556065707580-84 0. the age structure of a male population in a certain region on July 1. Example: In Table 7.Table 6.20405 0.01798 0.01142 0.01673 0. 1984.03372 0. x nqx x 25- sqx x sqx 0 5101520- 1- 0.52473 m. Because they reflect changes in mortality. 1989.00787 0. the probabilities of dying are the same as in Table 4 and the life expectancy at birth.14144 0. Suppose eo is increasing in half-year steps annually.39085 0.15 years. - 37 - . eo. Population Projection Model life tables are very important in population projection. For convenience. Calculate the age structure of the given population on July 1. is 72.00489 0. the influence of birth and migration are neglected here. is presented. suitable model tables may make population projection more accurate.00747 0.

According to information on death of the given population. Age group x Age group x p x Age group x PAge x group x P x P x 0-4 10-14 15-19 20-24 5-9 334241 269848 331553 357281 447889 25-29 30-34 35-39 40-44 45-49 471969 397317 276621+ 50-54 55-59 60-r:>4 196562 208190 65-69 70-74 208555 175050 130386 95594 73246 75-79 80-84 40743 18196 Step 1.38 - .Table 7. Table S shows its age-specific probabilities of dying and Table 8 presents person-years lived in each age group. Age group x 0-4 5-9 nx L Age group x nx L Age group x 50-54 nx L Age group x nx L 10-14 15-19 20-24 491235 487414 485954 484834 483235 25-29 30-34 35-39 40-44 45-49 481321 479152 476213 471762 464760 55-59 60-64 65-69 70-74 453427 433739 401702 356364 291300 75-79 80-84 85-89 225441 147410 93735 . Find the model table with eo = 72.15 in the family of pattern II for males. It was shown in section II that the second pattern for males is the best representation. Lx. a representative mortality pattern can be selected. the number of Table 8. Select the appropriate pattern in model life tables of China. Step 2.

83426 0.99399 0.67998 .5 = 74. values are reported as sioo in Table Table 9.94169 0. Age group x S2(x) Age group x S2(x) Age group S2(x) x Age group x S2(x) 10-14 15-19 20-24 0--4 5-9 0.95658 0.86349 0.69133 0.65387 0.92614 0.99525 0.99651 0.96111 50-54 55-59 60-64 65-59 70-74 0. The corresponding In Table 10 they are reported as S2(x).99392 0.97562 50-54 55-59 60-64 65-69 70-74 0.99770 0.99544 0.99065 0.99670 0.90971 0. x+5) to the next in 1984 can be obtained.96609 0.99834 0.99752 0.99277 0.99222 0. would be ratios can be + 2. The life expectancy 72. 1989.98516 0.88714 0.98851 0.99693 25-29 30-34 35-39 40-44 45-49 0. ratios from age interval (x.39 - .99700 0.From the formula: the survivorship Sex) = Lx+s/Lx These 9.65 in pattern II for males.588 Step 3.99784 0. Table 10.99604 25-29 30-34 35-39 40-44 45-49 0.75830 15-19 80-84 0. Age group x 1 S (x) Age group x Sl(x) Age group x S 1 ex) Age group x S 1 ex) 10-14 15-19 20-24 0-4 5-9 0. survivorship obtained similarly.79276 75-79 80-84 0.15 at birth of the given population on July 1.63.

The following is a reverse-survival method and illustrates the application of model life tables in indirect estimation . the age distribution of the given population on July 1. 85 The results are shown in Table 11.. 10.. . Age group x P 89 x Age group x p x 89 Age group x P 89 x Age group x p x 89 20-24 10-14 15-19 5-9 331936 269152 330697 356249 '25-29 30-34 35-39 45-49 40-44 446315 470069 395166 275140 193974 55-59 60-64 65-69 70-74 50-54 203686 200491 163482 117142 61147 80-84 75-79 56805 27404 IV.40 - . x+5) to the next age group in the period 1984-1989. Take S(x) =- i -(S l(x) + S2(x)) as the survivorship ratio of the given population from (x. Indirect Estimation Many methods of indirect estimation can be based on model life tables. Table 11. 1989. 15. can be calculated by using the following equation: sP89 x = sp84 x • S(x-5) x = 5. Then..Step 4.

15 and 74.01408.47 years. and the life expectancies at birth for males and females are 72.01367 and for females rw In 391620\01 n 3401807 = 0. The computational procedure is as follows: Step 1.Example: Suppose that: (1) in a certain region in 1984 there are N~ = 4038988 males and N~ = 3916202 females. Find a model table with life expectancy at birth of 72. (3) The mortality patterns of both sexes correspond to pattern II in model life tables of China.41 - . Then the following results can be obtained: . Estimate crude birth rates and sex ratios at birth for the periods 1974-1979 and 19791984. and their distributions by age group under 10 years are given as follows: Age group OA 334241 311519 5-9 269848 255814 (2) In 1974 there were 3522888 males and 3401807 females in the total population.15 years for males and another with eo of 74.47 years for females both in pattern II of model life tables for China. respectivel y. thus the average annual growth rate for males is rro In 4038988 10 In 3522888 = 0.

erm(S1.5) == 3523738 eO.42 - .5) = 3916202 = 3780750 .S-84) eO.01367(-7.5-84) = 4038988 eO. == N~ erm(76. Nr N~.5-84) = 3916202 = N~ • erw(Sl. are the number of person-years lived in age interval (x.Age group o 98964 99205 1-4 5-9 392271 394255 487414 490977 Here n~ and nL. erw(76.01408(-7. Calculate mid-period populations by sexes for 1974-1979 and 19791984. x+n) for males and females in the life table.5) = 3645411 N~ = N~. m m m Then: 5Lo == Lo + 4Ll == 98964 + 392271 == 491235 w w w SLo == Lo + 4Ll = 99203 + 394255 = 493458 Step 2.01367(-2.5) = 3903288 N7 = N~.0140S(-2.S-S4) == 4038988 eO.

... Estimate average annual births by sex in the mid-points of 1974-1979 and 1979-1984.Step 3. SNo w 10000 SLo = (493458) 311519 -100000 = 63130 - 43 - . . m 269848 Bl = ----.10000 = (487414) -100000 = 55363 SLS m SNS B2 w = -.

Step 4.78. The results are respectively: for 1974-1979: 1 1 55363+52103 CBRI =-N-m-+-N-w-· 1000% =3645411 + 3523738 1 1 sm + BW -1000% = 14.07% SR2 = -.99% sm SRI 100 = 106.44 - .1000% 2 2 2 = 3903288 68041 + 63130 + 3780750· 1000% = 17. W B2 Bm 2 100 = 107. Estimate crude birth rates and sex ratios at birth in the period 1974-1979 and 1979-1984. .26 = _1 • B1 B m 2 W for 1979-1984: + BW CBR2 = -N-m-+-N-w .

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->