INTRODUCTION
We noted toward the end of Chapter 1 that we demographers are more fortunate than many of our social science colleagues who must gather and develop their own databases. Generally speaking, most of the data we use have already been gathered for us, but not always. Indeed, some demogra-phers do gather their own data, especially those who use anthropological perspectives and who engage in ethnographic research (Greenhalgh,1990b, 1994; Riley,1998; Riley and McCarthy,2003). However, most of us use data already gathered and developed by other organizations. This chapter discusses the basic sources of demographic information, of which there are three.
The basic sources of demographic data are national censuses, regis-ters, and surveys. National censuses and registers differ in that the former are conducted on a decennial (or, in some countries, quinquennial) basis, while the latter, theoretically at least, are compiled continuously. Actually, registrationdata of population events are usually compiled and published annually or monthly, but they are gathered continuously. Acensusmay be likened to taking a snapshot of a population at one point in time, say, once every ten years, and in this snapshot getting a picture of the size of the popu-lation, its characteristics, and its spatial distribution. Conversely, a register may be thought of as a continuous compilation of major population events, often births, deaths, marriages, divorces, and sometimes migrations. As a birth or a death occurs, it is registered with the government; the registra-tions thus occur continuously.
Censuses and registers are intended to cover the entire population. In a national census, everyone in the population is supposed to be enumerated, and all the demographic events (births, deaths, and so forth) that occur in the population are supposed to be registered. Surveys, on the other hand, are by definition administered to only a fraction of the population. Yet they often gather data on many of the items included in censuses and registers, 15
plus additional items of interest to demographers not included in them. We now cover in some detail each of these three sources of demographic data.
NATIONAL CENSUSES
A national census is “the total process of collecting, compiling, and pub-lishing demographic, economic, and social data pertaining, at a specified time . . . , to all persons in a country or delimited territory” (United Nations, 1958: 3). The principal objective of a census is to obtain data about the size, composition, and distribution of the population. A typical census thus includes information about the size of the population and its social and geographic subpopulations, as well as data on their age and sex composi-tion and their educacomposi-tional composicomposi-tion (levels of literacy and educacomposi-tional attainment and extent of school attendance). Many censuses also contain information on economically active and inactive populations, including data on the industrial and occupational composition of the working pop-ulation, as well as economic (salary and income) data. Other population data in a typical census include information pertaining to country or area of birth, citizenship, language, recent migration experience, religion, and ethnicheritage, which refers to group distinctions based on shared cultural origins (Shryock, Siegel, and Associates,1976).
In the actual enumeration of the population, there are two ways to count people: by following a de jure method or by following a de facto method (Shryock,1964). In the case of a de jure enumeration, the census covers the entire territory of the country and counts persons according to their “usual” or “normal” place of residence in the country. A de facto enumeration also covers the entire territory of the country but counts each person in the population according to his/her geographical location on the day of the census undertaking. For instance, a person who resides with her family in Norfolk, Virginia, but who is traveling on census day and happens to be counted in College Station, Texas, would be counted as a resident of Norfolk if the census was a de jure census but would be assigned to College Station if it was a de facto census. Canada and the United States follow a de jure approach, as do many European countries, for example, Austria, Belgium, Croatia, the Czech Republic, Denmark, Germany, the Netherlands, Norway, Sweden, and Switzerland (United Nations, 1998).
The censuses of Colombia have been both de facto (in 1963 and 1973) and de jure (since 1985). Of the more than 230 countries conducting national censuses, however, the de facto type is much more common than the de jure (Wilmoth,2004: 65).
Shown inFigure 2.1is an example of a census questionnaire; this one was administered in the United States in the 2000 census. The questions on this instrument were usually answered by one person in everyhousehold
17 National Censuses
in the United States, and that person typically entered responses to each question for everyone residing in the household; these are known as the 100 percent questions. There were just a few such questions, and they dealt with age, sex, race, Hispanic origin, household relationship, and owner/renter status of the residence. Another much larger questionnaire containing many more questions, for example, dealing with education, occupation, income, mobility, and several other topics, was used in the 2000 census but was only administered to a sample of the population, roughly one in six households.
The census questionnaire containing the additional questions has come to be known as thelong-form questionnaire.
Census taking had its origins in ancient Egypt, China, and Rome, among other places, although only a few of these enumerations have sur-vived. There may have been a census conducted in China as early as 3000 BC, but demographic records for China and other countries for the very early periods no longer exist. Several census counts are mentioned in the Bible; one was undertaken at the time of the Exodus in 1491 BC, and another was conducted during King David’s era in 1017 BC. Roman cen-suses were conducted quinquennially for more than 800 years. The Romans extended the census enumeration to the entire Roman Empire in 5 BC, resulting in the popular biblical census story reported in St. Luke’s Gospel (Bryan,2004: 14).
It is difficult to determine when the first modern census was under-taken. Coverage was highly suspect in early efforts; women and children were seldom included. Censuses were often conducted to determine the fis-cal and military obligations of the citizens (Bryan,2004). Most countries of the world today conduct censuses. Some countries are late to census taking.
For instance, Chad and Oman did not take their first censuses until 1993.
Of all the countries in the world, only one, Lebanon, has never conducted an official population census. Most likely owing to the way the country was formed by the French, Lebanon has used national population and house-hold surveys for various enumeration estimates and has avoided, likely for political reasons, conducting an actual census (personal communication with Mary J. Chamie, July 10, 2007).
Of the more than 230 countries or areas in the world today, the United Nations reports that all but thirteen (Western Sahara, Guinea-Bissau, Liberia, Togo, Eritrea, Somalia, Democratic Republic of the Congo, Burundi, Angola, Myanmar, Afghanistan, Uzbekistan, and Bosnia and Herzegovina) conducted a national census in the 1993 to 2006 time period (United Nations,2007). Indeed more than 95 percent of the world’s pop-ulation has been counted in a national census conducted sometime during the decade of the 1990s (P. Johnson,2000).
Population censuses were conducted relatively early in the United States, starting with Virginia in 1624–1625. Various colonial censuses were
Figure 2.1.The 2000 U.S. Census of Population and Housing, 100% questionnaire.
conducted through 1767. In the United States, the principal reason and justification for conducting a decennial national census is to provide pop-ulation counts for the states of the country that are used to apportion the House of Representatives. The requirement for a decennial census was written in 1787 into Article 1, Section 2, of the U.S. Constitution as follows:
“Representatives and direct taxes shall be apportioned among the several states which may be included within this Union according to their respective
19 National Censuses
Figure 2.1(continued).
numbers. . . . The actual enumeration shall be made within three years after the first meeting of the Congress of the United States, and within every subsequent term of ten years in such manner as they shall by law direct.”
The first national census was conducted in 1790, and one has been conducted every ten years since that time. The 1790 U.S. census, however, counted people only according to the following categories: 1) free white males 16 years and over, 2) free white males under 16 years of age, 3) free white females, 4) slaves, and 5) other persons, that is, persons not included
in the first four categories. White females were not counted by age, and nonwhite people were counted neither by age nor by sex. Compare these restrictions with the much more inclusive questions asked in the 2000 U.S.
census (seeFigure 2.1).
Today, censuses and census data are very important for the functioning of government bodies. Box 2.1 shows exactly how the U.S. House was apportioned using data from the 2000 Census (Baumle and Poston,2004).
Censuses are quite expensive to conduct; the cost of the 2000 U.S.
census exceeded 4.5 billion dollars (Gauthier,2002). But census data pro-vide government officials with useful and necessary information about the people in their country. Governments use census data in virtually all fea-tures of public policy, for example, how many children the public schools need to serve and where to place new roads. Census results also provide the denominator data for crime rates, death rates, per capita income figures, and other statistics that are needed to administer local and national gov-ernments. Private businesses require census data for their market analyses and advertising activities (M. Anderson, 2003). Many demographers and other social scientists use census data to test their theories and conduct their analyses.
For instance, one of the questions in the 2000 census asked everyone living in a household with two or more persons about their relationship with the person who is known as the “householder.” The householder is meant to be “the member of the household in whose name the home is owned, being bought or rented” (Barrett, 1994: 16). Operationally, it refers to the person taking the major responsibility for filling out the census form. Look at question #2 in the second part ofFigure 2.1for the actual wording of the householder relationship question.
One of the responses to the householder relationship question is
“unmarried partner.” This response is used to identify persons in the house-hold who are not related to the househouse-holder but who have a “marriage-like” relationship with the householder. Census procedures permitted
BOX 2.1 USING CENSUS DATA TO APPORTION THE U.S.
HOUSE OF REPRESENTATIVES
The major objective in apportioning the U.S. House of Representatives is to assign equitably the 435 seats to the fifty states (the District of Columbia is not included in the apportionment and, thus, does not receive representation in the House). There are several constraints: 1) The total number of House seats must equal 435; 2) partial represen-tatives cannot be assigned to states, nor can represenrepresen-tatives be given fractional votes; 3) representatives may not be shared by two or more states; and 4) every state must be assigned at least one seat in the House.
21 National Censuses
The first fifty seats are automatically assigned, one per state; the purpose of the apportionment method is to divide up the remaining 385 seats. The apportionment method of Equal Proportions indicates which states should receive second seats, which states should receive third seats, and so forth. The U.S. Constitution does not provide instruc-tions on how apportionment should be carried out, but the underlying assumption is “one man, one vote.” That is, no one person should have more of a voice than another person. As a result, representatives are assigned from states in proportion to their populations. The method of Equal Proportions was first used to apportion the House in 1940 and has been used ever since. It is a divisor method that first develops a targetratioof population to representatives that is based on data for the nation. In 2000, the apportionment population (the population counted by the Census Bureau residing in each state plus certain individuals living overseas who claim the state as their “state of residence,” namely, mil-itary personnel and U.S. government employees and their dependents) of the United States was 281,424,177. Hence, the target ratio in 2000 was 646,952.1 (or 281,424,177 divided by 435). This ratio, also called a divisor, is then divided into the apportionment populations of each of the states to obtain quotients. The method of Equal Proportions endeav-ors to ensure that “the difference between the representation of any two states is the smallest possible when measured both by the relative dif-ference in the average population per district, and also by the relative difference in the individual share in a representative” (Schmeckebier, 1941: 22). The method gives to a state another representative “when its [apportionment] population, divided by the geometric mean of its present assignment of representatives and of its next higher assignment, is greater than the [apportionment] population of any other state divided by the geometric mean of the assignment to such other state and its next higher assignment” (Schmeckebier,1941: 22).
The first step in using the method of Equal Proportions is to multiply the apportionment population of each state by the following fraction:
1
N(N−1)
where N equals the particular seat being claimed, that is, the second seat or the third seat or the fourth seat, and so on. This provides numbers known as priority values. For instance, the proportion used in determin-ing a state’s claim to a second seat is:
1
2(2−1) = 1
√2 = 1
1.41421356 =0.70710678
The proportion used in determining a state’s claim to a third seat is:
1
3(3−1) = 1
√6 = 1
2.44948974 =0.40824829
The rounding rule for this method is to round a state’s quotient either up or down, “depending on whether or not the quotient exceeds the ‘geometric mean’ of these two choices” (Balinski and Young,1982:
62). The geometric mean of two numbers is the square root of their product. Thus, according to the method of Equal Proportions, if a state had a quotient of 1.39, it would receive one representative because the geometric mean of 1 and 2 is 1.41; however, if a state had a quotient of 1.42, it would receive two representatives.
In the actual apportionment calculations, the rule per se need not be invoked. Instead, one relies entirely on the proportions developed for the various seats. Thus, once the proportions are developed for determining the priorities for the various seats (we have shown the proportions for seats 2 and 3), they are multiplied by the apportionment populations of each of the fifty states. That is, the proportion used for determining the states’ priorities for a second seat (0.70710678) is successively multiplied by the apportionment populations of each of the fifty states; this proce-dure is then repeated using the proportion to determine the states’ pri-orities for a third seat (0.40824829) and so forth. After all of these mul-tiplications have been completed, the resulting priority values are then ranked in order, the largest first and the smallest last. The 385 House seats are assigned to the states with the 385 highest priority values.
In the following table, we report the application of the Method of Equal Proportions in 2000 and identify the states receiving the first six seats and those receiving the last six seats. We also show the states that would have received the three seats beyond the 435th seat if more than 435 seats were assigned. In the 2000 apportionment, California received the 51st seat. Its priority value for a second seat, 23,992,697, was obtained by multiplying its 2000 apportionment pop-ulation of 33,930,798 by the “second seat” proportion of 0.70710678.
Texas received the 52nd seat with its priority value for a second seat of 14,781,356, which was determined by multiplying its 2000 appor-tionment population of 20,903,994 by 0.70710678. The 51st and 52nd seats were thus assigned to the two largest states, California and Texas.
New York was the third largest state in 2000, but New York did not receive the 53rd seat because its priority value for a second seat of 13,438,545 was smaller than California’s priority value for a third seat of 13,852,190 (the priority value for California’s third seat is obtained
23 National Censuses
by multiplying California’s apportionment population of 33,930,798 by the “third seat” proportion of 0.40824829). So California received the 53rd seat and New York the 54th seat. Florida received the 55th seat as its second seat, and California received the 56th seat as its fourth seat.
The table also shows the states receiving the last six seats in the House, the 430th through the 435th seats. Note, for instance, that Georgia’s priority value for a 13th seat was slightly larger than Iowa’s claim for a 5th seat, so that the 430th seat was assigned to Georgia and the 431st to Iowa. North Carolina received the 435th and last House seat allocated as its 13th seat. The states of Utah, New York, and Texas were next in line to receive the 436th, 437th, and 438th seats had the House allocated three more seats. We have estimated the pop-ulations that would have been needed for either Utah or New York or Texas to have been allocated North Carolina’s 435th seat. If no other state’s population changed, Utah would have needed an apportionment population in 2000 of 2,237,574, which is a mere 860 more persons than its actual 2000 apportionment population. New York would have needed another 47,284 persons in its 2000 apportionment population and Texas another 86,312 persons for either state to have received the 435th seat (Baumle and Poston,2004).
Application in 2000 of the Method of Equal Proportions: Allocating the first six and last few seats
Numbered
seat in House State
Numbered seat in the State
Priority value First six seats
51 California 2 23,992,697
52 Texas 2 14,781,356
53 California 3 13,852,190
54 New York 2 13,438,545
55 Florida 2 11,334,137
56 California 4 9,794,978
Last six seats
430 Georgia 13 657,084
431 Iowa 5 655,598
432 Florida 25 654,377
433 Ohio 18 650,239
434 California 53 646,330
435 North Carolina 13 645,931
Three seats beyond the 435th
436 Utah 4 645,684
437 New York 30 644,329
438 Texas 33 643,276
respondents to check “unmarried partner” whether or not the person’s sex is the same as that of the householder. It is thus possible to identify the number of adults in the Unites States who are unmarried partners with persons of the same sex and then calculate the numbers of same-sex adult males and same-sex adult females who are living together. Because this response is meant to reflect a marriage-like relationship between the two persons, demographers make the assumption that these data on same-sex households (male-male or female-female) represent households inhabited by partnered gay men or partnered lesbians (Baumle, Compton, and Poston, 2009; D. Black et al.,2000).
One study used these same-sex data from the 2000 census and cal-culated gay male partnering rates and lesbian partnering rates for the 331 metropolitan areas of the United States (Baumle, Compton, and Poston, 2009). The authors showed that the gay male rate has a mean value of 20.0, meaning that across the 331 metropolitan areas there are, onaverage, 20 gay male cohabiters for every 1,000 never-married males of age 18 and older. (In standard usage, an average is the one value that best represents all cases in a set.) San Francisco has the highest value with a score of almost 61. San Francisco contains the Castro Valley neighborhood, a well-known gay male enclave, making the high prevalence of partnered gay males in San Francisco not a surprise. Dubuque, Iowa, has the lowest score, of about 6 gay male cohabiters per 1,000 never-married males. Dubuque has strong links with the Catholic Church, including the presence of a number of monasteries and motherhouses and two Catholic universities. This strong historical tie with Catholicism may well be linked, at least in part, to the low presence of same-sex male partners in the city, owing to the church’s stance against homosexual conduct and gay marriage.
For partnered lesbians living in metropolitan areas, Amanda Baumle and her colleagues (2009) reported an average prevalence rate of almost 27. The Santa Rosa, California, metropolitan statistical area (MSA) has the highest value, a score of more than 72; for every 1,000 never-married women of age 18 and older in the Santa Rosa MSA, there were almost 72 lesbian cohabiters. The Santa Rosa MSA is comprised of a single county bordering the Pacific Ocean, Sonoma County, and is immediately north of Marin County and San Francisco. Its proximity to San Francisco, along with a somewhat more rural locale, perhaps contributes to its high-partnered les-bian prevalence score. The Provo–Orem, Utah, metropolitan area has the lowest score, at 9 per 1,000. Nearly 90 percent of Provo’s population is Mormon (Hamby,2005). Also, Provo is home to Brigham Young Univer-sity, a large private university that is operated by the Church of Jesus Christ of Latter Day Saints. Its adherents oppose marriages of gay males and of lesbians, and they proscribe homosexual behavior in general. Perhaps as a
25 Registration Systems
result, gay men and lesbians in Utah have been the subject of a great deal of litigation and restrictive legislation (Hamby,2005).
Baumle and her colleagues (2009) also found that for the most part, the gay male rates tend to vary in the same way as the lesbian rates. Metropoli-tan areas with high rates of gay male partnering have high rates of lesbian partnering, and areas with low gay male rates have low lesbian rates. But most of the metropolitan areas, 305 of the 331, have higher lesbian rates than gay male rates. The authors suggest that partnered gay men appar-ently have a few favorite destinations, including San Francisco, Atlanta, Los Angeles–Long Beach, Miami, Jersey City, Washington, D.C., New York, and Fort Lauderdale, where their prevalence rates surpass those of part-nered lesbians. Partpart-nered lesbians, conversely, are concentrated more than are partnered gay men in metropolitan areas in general, tending not to prefer particular areas to the degree that gay men prefer them (Baumle, Compton, and Poston,2009).
This is but one example of the many and different kinds of demo-graphic research questions that may be answered with data from censuses.
We turn next to a discussion of the second source of demographic data, registration systems.
REGISTRATION SYSTEMS
Whereas censuses provide a cross-sectional (one point in time) portrayal of the size, composition, and distribution of the population, registration systems pertain to the population’s demographic events (births and deaths and, in some places, migrations) and measure them as they occur. While censuses are static, registers are dynamic and continuous. Registers apply principally to births and deaths, although many countries also maintain reg-istrations of marriages, divorces, and abortions. Some countries maintain a migration registration system.
Strictly speaking, as Lars Ostby (2003: 763) has noted, apopulation registeris a list (i.e., a register) of persons that includes the name, address, date of birth, and a personal identification number. Some registers have been maintained for centuries, such as those in church parishes that record the baptisms and the deaths of the parishioners. In Europe, the Nordic countries and the Netherlands maintain some kind of population register, and many developing countries either have them in place or are planning to implement them. In Eastern Europe under the Communists, “population registers were used for control (of the people) as well as for administrative purposes, and the successor regimes for the most part have not maintained them” (Ostby,2003: 763). The United States does not maintain any kind of national population register.
The earliest example on record of a population register of families and related household events was in China during the Han Dynasty (205 BC–AD 220). Indeed, as Irene Taeuber (1959: 261) noted, a special demo-graphic tradition of China and the East Asian region as a whole was pop-ulation registration. Its major function, however, “was the control of the population at the local level” (Bryan, 2004: 25) and not necessarily the collection of continuous data on demographic events.
Population registers are of interest to demographers because they con-tain birth and death records (certificates). But not all birth and death reg-istrations occur in the context of population registers. In fact, since a large number of countries do not maintain them, the registration of many births and deaths occurs outside population registers.
For most countries in the world, the recording of vital events, that is, births and deaths along with marriages, divorces, fetal deaths (stillbirths), and induced termination of pregnancies (abortions), are recorded in their civil registration systems. But these registration systems need not necessarily be population registers. Indeed, many are not. Although civil registration data are not 100 percent accurate and complete in the more developed nations, their quality is far better than that in the poorer nations. John Cleland (1996: 435) has observed that although civil registration systems in developing countries are “seriously defective, it would not be correct that the data are of little value to demographers.” Demographers have developed special techniques for data adjustment and analysis, yielding a rough notion of trends and differentials in these demographic events (Popoff and Judson, 2004).
As articulated by Mary Ann Freedman and James A. Weed (2003:
960), “Vital statistics form the basis of fundamental demographic and epidemiologic measures.” Vital statistics are the data derived from civil registration systems, as well as from the actual records of vital events. The modern origin of vital statistics and their registration may be traced to the English ordinance in 1532 requiring that parish clerks in London maintain, on a weekly basis, the registration of deaths and christenings (Bryan,2004:
25). These reports were begun in response to the plagues of the late sixteenth and early seventeenth centuries and were published in a nearly unbroken series for decades. Merchants used those data as a rough gauge of the likelihood of their clientele to flee to the countryside during epidemics (Kraeger,1988: 129). John Graunt’s ([1662] 1939)Bills of Mortalityis a well-known demographic analysis of these data (seeBox 2.2).
With regard to the modern era, Simon Szreter (2007) has written that the registrations of one’s birth and death are fundamental human rights.
The second clause of Article 24 of the International Covenant on Civil and Political Rights (ICCPR) of the United Nations states that “every child
27 Registration Systems
BOX 2.2 JOHN GRAUNT
John Graunt is deemed by many (Bogue,1969: 9; Poston,2006a: 254) to be the founder of demography. He was born in London in 1620, raised as a Puritan, and later in life became a Catholic. He died in London in poverty in 1674. Although lacking any higher education and untrained in the sciences or mathematics, he published in 1662 the first-known quantitative analysis of a human population,Natural and Political Observations Made Upon the Bills of Mortality.
The “Bills of Mortality” were weekly accountings and reports of the London parish clerks of all the deaths and christenings. These reports were started in response to the plagues of the late sixteenth and early seventeenth centuries and were published in a nearly unbroken series for decades. Merchants used data from the Bills as a rough gauge of the likelihood of their clientele to flee to the countryside during epidemics (Kraeger, 1988: 129). Graunt studied this mass of data searching for regularities. He is credited for being the first to recognize that more males are born than females and that females have greater life expectation than males. He also was one of the first to recognize the phenomenon of rural to urban migration. He also developed a crude mortality table that eventually led to the modern life table; as shown in Chapter 5 of this book, life tables are the basis for calculating life expectancy. Graunt also set a precedent for one of demography’s oldest traditions, namely, the evaluation of data “to learn the extent, types, and probable causes of errors” (Bogue,1969: 9). He “carefully evaluated the bills for their numerical consistency and reliability of compilation, and presented his evidence at length so that his readers might judge it independently”
(Kraeger,1988: 129).
Although Graunt died in obscurity, his lasting monument is his Natural and Political Observations, a book that to this day is a joy to read (Poston,2006a).
shall be registered immediately after birth and shall have a name” (Szreter, 2007: 67). The ICCPR also states that “for nation states to take appropriate measures to protect and enhance the life expectancy of their populations, they must have at their disposal accurate and detailed information about patterns and trends of mortality” (Szreter,2007: 68), thus also requiring death registration. (Life expectancy is the average number of years yet to be lived by people attaining a given age, according to a given demographic table.)
How complete is the registration of births and deaths in the world today? For the year 2000, the United Nations (UN) International Children’s
Emergency Fund (UNICEF) Research Center has estimated that there were around 50 million babies unregistered, which is more than two fifths of all the babies born in 2000 (UNICEF, 2001). The unregistered children are often found in countries where “there is little awareness of the value of birth registration, where there are no public campaigns, where the registra-tion network is inadequate, or where the costs of registraregistra-tion of children are prohibitive” (UNICEF, 2002: 10). In general, most unregistered babies are born in developing nations, largely because these countries are more likely to face political, administrative, and economic barriers to registration.
In some countries, gender discrimination and son preference also lead to female babies being excluded from the birth registration (Hudson and den Boer,2004). UNICEF has noted that in the year 2000, more than 70 per-cent and 63 perper-cent of births in sub-Saharan Africa and Asia, respectively, were unregistered. In South Asia alone, there were an estimated 22.5 mil-lion unregistered births, the largest number among all the areas of the world.
This does not mean, however, that all developing countries have seri-ously incomplete birth registration. Many countries in the former Soviet Union have virtually universal coverage of births. This is due likely to their well-established birth registration systems, good medical facilities, and well-trained medical personnel.
Regarding deaths, we do not know as much about the completeness and coverage of death registration around the world. Like the situation with birth registration, incomplete death registration occurs more often in developing nations. For example, only 57 percent of infant deaths were registered in Egypt in the early 1990s (Becker et al.,1996).
The registration of births, marriages, and deaths in the United States began with registration laws in Virginia in 1632 and later in other colonies.
We noted earlier that the U.S. Constitution provides the requirement for a decennial census; but there is no such federal requirement for a national vital registration system. Legal authority for the registration of vital events in the United States lies with the individual states. The first U.S. census was conducted in 1790, but the complete coverage of births and deaths occurred much later.
We noted earlier that in seventeenth-century England, the registra-tion and maintenance of baptism, marriage, and burial records were the responsibilities of the clergy. This practice was also followed by the English colonies in North America. In 1639, courts in the Massachusetts Bay Colony declared that birth, death, and marriage reporting would be part of their administrative system. Thomas Bryan has written that the Massachusetts Bay Colony “may have been the first state in the Western world in which maintaining such records was a function of officers of the