The most serious criticism of our earlier proposal was the choice of Colombia as a research site instead of some other Latin American nation. Site-selection is of utmost importance to the success of what we hope will be a prototype for the region, and indeed other regions around the globe. Although we presented only three reasons for choosing Colombia, as a matter of fact, we arrived at our decision only after consultation with demographers and census authorities in all but one of the larger countries of Latin America (those with more than twenty million inhabitants in the mid-1990s--Brazil, Mexico, Colombia, Argentina, and Peru, but not Venezuela) and four smaller ones (Chile, Ecuador, Paraguay, and Costa Rica). Brazil was quickly dropped from consideration because its scale of difficulty is substantially greater than for any other Latin American country, while our Brazilian expertise is substantially less. A Spanish-speaking country serves as a more useful prototype.

Issues of data quality, scientific salience, and chronological continuity were so fundamental to our decision that we did not discuss these in our initial proposal, but we take advantage of this re-submission to do so now.

Data quality. We sought a country where post-enumeration surveys are common and coverage rates 85% or higher. There is widespread agreement that on the whole Colombian censuses are of good quality, approaching the highest standards in Latin America, such as those of Argentina, Costa Rica, and Chile (Dirk Jaspers, Susan De Vos and Joe Potter, private communications). Colombia is one of the few countries in Latin America where both pre- and post-enumeration surveys are regularly carried out. Beginning with the census of 1973, the Colombian statistical agency, DANE, has conducted—and published—post-enumeration surveys. Under-enumeration rates for Colombia fall in a range of 11.5-15% (Potter and Ordoñez, 1976; DANE 1990, 1996). While political violence, as noted by one reviewer, certainly affects enumeration coverage and quality, this problem is limited primarily to peripheral regions. Although it is true that Colombia does not rank among the best countries in the world in terms of census coverage, its record is superior to that of most Latin American countries, including Ecuador. Initially, Ecuador was a favored site for this project, and, indeed, the PI’s interview with officials of the Instituto Nacional de Estadística y Censos in July 1997 revealed unbridled enthusiasm for our project. However, INEC’s spare archives, disorganized library, scarce resources, and paucity of studies evaluating census procedures or undercount persuaded us that Ecuador was not the most auspicious site for developing our prototype. A comparison of Ecuadorian and Colombian census publications reveals further qualitative differences, to the advantage of Colombia. Where INEC publishes Ecuadorian census figures with little discussion, explanation or qualification, DANE publications demonstrate much technical sophistication and great concern with census quality, interpretation and inference. The DANE archives are impressive in terms of their organization, accessibility, and completeness. For example, a query in the computerized catalogue of the central DANE archive for documentation pertaining to the census of 1973 yields 298 reports totaling more than 5000 pages.

Scientific salience. Both qualitative and quantitative indicators point to Colombia as a top-ranked site for this project. If the ethnic and ecological diversity of Colombia makes it an exceedingly interesting laboratory for studying compelling scientific issues of late twentieth-century Latin America (see below), quantitative indicators, measured by citations in Population Index, place Colombia near the top as well. In scholarship, Colombia ranks fourth in Latin America, behind Mexico, Brazil and Peru in a cluster with Argentina, Costa Rica and Cuba, but well ahead of Chile, Ecuador and other Latin American candidates. We resisted the allures of Chile and Costa Rica as countries that are too small, and Argentina as too well-studied and relatively less interesting due to its more homogeneous, wealthier, Europeanized population.

Chronological continuity of available microdata (at least one census every decade since the 1960s). Mexico, our long-held preference, ultimately failed the continuity test. In vain we searched computer data libraries throughout the Western Hemisphere to locate microdata for the 1980 census of Mexico. Finally, we were forced to accept the often-repeated dictum of our advisers that no copies of the 1980 tapes survived the 1985 earthquake. Then too, for 1960, INEGI officials, in a misguided attempt to anonymize the microdata, deliberately obliterated household and family structure. We were forced to conclude that, as a prototype, Mexico is not as well suited as Colombia. Nevertheless, because of the large Mexican-born population in the United States, we plan to undertake an integration of Mexican censuses in the near future.

Colombia satisfies the criteria of quality, salience, and continuity as well as other less tangible considerations. As noted below, the eagerness of the Colombian statistical authorities and academics to collaborate in the undertaking was an important factor (see the attached notarized agreement, dated Dec. 4, 1998, between the University of Minnesota, DANE, Universidad Externado and Universidad Nacional) as well as the notion that this project might assist a people determined to reconstruct a country troubled by guerrilla warfare and criminality. Finally, it should be noted that we see Colombia as only a beginning. If this project is successful, over the next decade we expect to extend the Col-IPUMS paradigm to Canada and Mexico as well as other countries in Latin America and beyond.

The third set of reviewer criticisms relates to the task--its magnitude, our ability to complete it and a concern with duplicating what CELADE may have already accomplished. To improve our plan, indeed to return to a more ambitious model of international collaboration which our previous proposal sacrificed for the sake of economy, we propose to expand our pool of consultants as well as extend our time in the field.

First, regarding CELADE, we seek to capitalize on its long history of using Latin American censuses, including those of Colombia, by working at CELADE for approximately one month at the beginning of this grant, and by drawing on the expertise of a CELADE consultant for several weeks each year of the grant (see the attached letter from Dirk Jaspers). CELADE has confirmed that its copy of the 1964 Colombian census microdata sample is recoverable. The availability of the 1964 census will permit a significant extension of the Col-IPUMS project, at modest cost. Dr. Jaspers also assures us that Col-IPUMS will not duplicate CELADE’s long abandoned effort at integrating Latin American censuses (OMUECE). That effort failed after the 1970 round of censuses due to the high computing costs in those years and CELADE’s lack of commitment to continuing the project into the 1980s.

Second, as the reviewers helpfully noted, the magnitude of the proposed integration is indeed great and as with any large-scale, novel enterprise there are risks that it may not succeed. However, the Col-IPUMS effort is at least an order of magnitude less than what we have already accomplished with the U.S.-IPUMS, 1850-1990. The need to revise our proposal provides the opportunity to improve the odds of success. The PI, will work for five months on site in the DANE archives and library in Bogotá beginning July 1, 1999, thanks to a recently awarded Fulbright Fellowship. During that time, our Colombian partners have agreed to host a seminar/workshop on the integration of Colombian censuses with the PI, in-country consultants, and other Colombian specialists as participants. This will respond to one reviewer’s advice that the PI allocate additional research time in-country, but without additional cost to the project.

For reasons of economy our initial plan proposed to use consultants sparingly. Now, we propose to form a partnership with two distinguished Colombian social scientists who have long worked with Colombian microdata. They and their collaborators will not only contribute their many years of expertise to the project but they will also become Col-IPUMS intellectual co-authors assuring widespread acceptance—by themselves, their students and their colleagues. Although usage is not expected to be a problem (US-IPUMS transmits 120 gb of US data per month), we will promote the use of Col-IPUMS by inviting a select group of national and international scholars to a data-release conference at the University of Minnesota near the conclusion of the project. Finally, CELADE will mirror the Col-IPUMS site for Latin American users or others, should over-loading become a problem at Minnesota.

This project seeks funding to create Public Use Microdata Samples (PUMS) for the Colombian censuses of 1964, 1973, 1985, 1993 and 2001. Microdata from the Colombian censuses has never been made publicly available, but the original data used to create the aggregate tabulations for these census years do survive in machine-readable form. Thus, it will be possible to create useable PUMS files and associated documentation for Colombia at modest cost. We plan to model the Colombian samples on the Integrated Public Use Microdata Series (IPUMS) created at the University of Minnesota, and to the extent feasible to make the coding schemes entirely compatible with the existing U.S. census microdata (Ruggles and Menard 1995). These scientifically drawn samples, once integrated into a single databank with uniform codes and documentation and released into the public domain, will constitute the richest source of quantitative information on long-term change for any Latin American population. The integrated Colombian PUMS will be distributed over the Internet using the system developed at the University of Minnesota for United States census samples (Ruggles, Sobek and Gardner 1996; Sobek and Ruggles 1999).

This project is a collaboration between the University of Minnesota and the Departamento Administrativo Nacional de Estadística (DANE), the Colombian census authority. DANE, with assistance from CELADE in the case of the 1964 census, is providing the machine-readable raw data files for each census year. We are contracting with DANE for several technical aspects of the project to tap the expertise of DANE cartographers, systems analysts, and statisticians. The project will also gain priority access to the 2001 Colombian census. We will be able to plan our integration using the 2001 census schedule, technical definitions, and administrative boundaries as the standard.

The work to be carried out at the University of Minnesota will involve the following steps:

In the long run, the investigators envision the creation of a single integrated series composed of public use samples of recent censuses for the larger Latin American countries for which the microdata still exist. For example, Mexico, the obvious first choice for the IPUMS prototype, was rejected for the initial trial, because the 1980 census tapes were destroyed in an earthquake and the surviving 1960 tapes, while readable, suffer a serious technical defect—the destruction of household ordering. The principal investigator possesses samples for 1960, 1970 and 1990. Mexican officials are now eager to develop an integration agreement—thanks in part, to the knowledge that the Col-IPUMS project will soon be underway. The pilot project for Colombia will establish standards and procedures for a wide-range of the most common variables found in Latin American censuses. Solutions to inconsistencies in the Colombian series will be designed to anticipate future difficulties with censuses for other Latin American countries.

Social scientists have increasingly recognized the need to study society as a process in time. If we confine our analyses to the state of society at a single moment, we cannot hope to understand the sources of social change. Sociologists, demographers, economists and public health analysts have developed a variety of quantitative data sources to study social change, including retrospective surveys, and longitudinal surveys. Although such data sources are essential, their analytical usefulness is limited.

Censuses are the most consistent general source of information about populations throughout the Americas, and Colombia is no exception. Quantitative studies of change in health, mortality and demographic processes in the broader society have always relied on published tabulations of statistical data, but these have substantial limitations. Census officials only attempted the most rudimentary cross-classifications of their data, and much of the information collected was never tabulated at all. Morever, aggregate tabulations in published census documents are poorly suited to multivariate analysis.

In the United States, the Census Bureau has addressed the limitations of tabular data by producing individual-level public use microdata samples for each census year since 1960. In Latin America, the United Nations sponsored Centro Latinoamericano de Demografía (CELADE) encouraged Latin American countries to construct similar samples or provide original census tapes to CELADE for that purpose. Many samples were so drawn, but their use was restricted to CELADE researchers who made only limited efforts to integrate the various censuses either chronologically (OMUECE, for censuses carried out between 1960 and 1976) or spatially (13 countries represented by two censuses, five by only one, and seventeen by none). Nevertheless, these census samples proved to be a valuable resource, since they allowed CELADE researchers to design sophisticated multivariate models tailored to their specific research questions.

In the United States, public use samples have proven their worth in studies of social change, because they substantially reduce the problems of incompatibility in the published data for different census years. In addition, the public use samples have allowed researchers to move beyond simple tabular analysis and apply increasingly complex, nuanced multivariate techniques. The existence of these data have significantly expanded the power of quantitative social science research.

In Latin America, the potential impact of public microdata samples are even greater because the CELADE census samples were never released to the public. In place of CELADE’s partial, cross-national integration of selected variables, with scant documentation, and jealously guarded dissemination, we propose a total integration of all available microdata for a single country with full documentation and open public-access. It must be noted that CELADE’s recently developed whole-census system REDATAM (implemented by Dr. Jaspers) is no substitute or competitor for the IPUMS paradigm. For reasons of confidentiality, REDATAM offers a powerful analytical engine, but cannot allow researchers to gain access to entire census data files for individuals with detailed geographic encoding. Planners will appreciate the finely grained statistics for small places provided by REDATAM, just as scientists will favor the analytical flexibility of stratified samples disseminated by Col-IPUMS.

The original census data tapes included a wide range of information—much of it concerned with public health, such as the availability of sanitation services, source of water supplies, type of cooking fuel, or housing construction materials—that was rarely studied (De Vos and Arias 1996). Coupled with responses to questions on fertility and mortality, these data offer exceptional opportunities for pinpointing the correlates of mortality change (or the lack of them) at the local, regional and national levels.

In Colombia, the availability of original census tapes means that it is still possible to construct public use microdata samples, using the most advanced specifications currently devised. This is a major undertaking, made more difficult by the fact that each census has a different format, different coding schemes, and different documentation. Although all Colombian censuses are remarkably similar in terms of questions asked and responses permitted, standardizing them on a uniform set of codes is a major undertaking. In fact, the only variable that is readily comparable across census years is age, and even there the censuses differ in treatment of missing, illegible, and inconsistent data. Even sex is not uniformly coded in the Colombian case. As the number of valid options increases, so does the confusion caused by inconsistent coding schemes. Published documentation for the censuses is minimal by international standards, and much essential information exists solely in the memories of DANE technical staff or remains unpublished in DANE archives. The limited published documentation is for the most part organized differently from one census to another, and the treatment of comparability issues is often cursory.

Even if the Colombian census files were readily available to researchers, their incompatibility in their present form would mean that multi-sample studies would require a large initial investment to prepare the data for use. In contrast, in the United States, thanks to the ready availability of the Census Bureau PUMS and their subsequent integration by the Historical Census Projects, there is an outpouring of valuable research. In Latin America, only a few researchers have been able to obtain census microdata through individual arrangements with national statistical offices. Such raw data are difficult to use because of inadequate documentation. In the rare instances in which researchers have attempted to use more than one census year simultaneously, they face daunting comparability problems. Given the complexity of the datafiles and the often-subtle differences among them, the potential for error is large.

This proposal seeks funding to transform Colombian census data into public use microdata samples for 1964, 1973, 1985, 1993, and 2001; integrate the resulting samples into a single consistent format; and prepare documentation oriented to the use of the samples as a series. In the long run, we anticipate expanding this prototype to other larger Latin American countries, such as Mexico (McCaa 1997). In the meantime, the Colombian integrated microdata series will provide compatible individual-level data for a large stratified sample of the Colombian population in five census years spanning four decades. The series will constitute a resource of great utility for the study of long-term social change in Colombia, and a model for other Latin American countries. Even should the Minnesota paradigm for integrating census microdata not catch on elsewhere, the Colombian databank will still be of great value for understanding the dramatic transformations occurring in Latin America at the end of the twentieth century.

In the case of the American microdata files, concerns about confidentiality have always been an issue of great concern. We will adopt similar procedures for the Colombian data. To respect Colombian confidentiality regulations (articles 74 and 75, decree law 1633 of 1960), we will eliminate all references to name, addresses, and geographic location below the district level. In addition, we will apply appropriate topcodes to variables such as income to ensure that individuals cannot be identified. Our challenge is to attain the highest possible sampling density, with the greatest possible geographical detail, without compromising confidentiality. Our proposed ten percent sample density is conservative. DANE will approve the highest density compatible with confidentiality safeguards, but actual testing will be required to arrive at that figure.

Colombia was chosen as the site for the initial trial for three reasons. First, the past quarter-century's machine-readable census tapes with a wide range of demographic variables still exist. Second, Colombian researchers and the Colombian National Statistical Bureau readily agreed to collaborate in the PUMS project. Third, the Colombian data offer extraordinary research opportunities for a number of contemporary policy issues—topics such as the demography of political violence (the most important public health issue in Colombia in recent decades), the social correlates of violence-related physical disabilities, mortality differentials by region and race, immigration to and from the United States, movement of women into the workforce, the spread of public education, urbanization, fertility transition, and the rise of the nuclear family.

The range of potential topics that can be addressed with these data is far too great to describe within the page limitations of a grant proposal. The following paragraphs are intended only to suggest some of the most obvious topics of investigation.

1. Women in the Workforce. The place of women in the workforce is currently one of the most controversial subjects in the study of Latin American women and gender issues more broadly. Some critics would deny any validity at all to Latin American censuses regarding women's labor. They argue that questions on work were designed with males in mind, based on an advanced economy model where jobs were stable, hours standardized, tasks routinized, and work calendars unvarying (Wainerman and Recchini de Lattes 1981; Gomez 1981; León 1985; Aguilar 1985; Bustos and Palacios 1994; Safa 1994). While such criticisms are justified to some extent, solutions to many of these objections may be found in the multiple questions on work presently available in microdata census samples, not only in the case of the United States (Sobek 1997), but also for many Latin American countries, including Colombia (McCaa 1998).

The principal investigator's analysis of the Colombian census microdata for 1973 and 1985 based on an ad hoc "integration" reveals a twenty percentage-point increase in formal labor force participation for women in only twelve years. The pattern is particularly striking for married women, for whom one factor alone—higher levels of education—accounts for over half the increase (results derived from a rough integration of a small set of key variables for two censuses with no regional or geographical detail). Comparing these results with IPUMS-derived data for the United States from 1880 to 1990 shows that by 1985 Colombian married women had attained participation patterns by age that closely paralleled those for married women in the United States in 1970. In both cases some 40% of married women aged twenty-five through forty worked in the formal labor force.

In the United States, over the three decades from 1940 to 1970, a great transformation occurred in the rate of married women in the workforce. In Colombia, only twelve years were required for a similar change to occur. Even more surprising in the Colombian case is that after education the second most important factor for explaining change for married women was not declining fertility, poverty (using an index based on public services available to the household), or even spouse's position in the workforce, but rather the husband's own educational attainments. A logistic regression model of these variables reveals that the greater a husband's education, the more likely that his wife worked in the formal labor force—even after taking into account her own level of schooling (McCaa 1998).

Microdata on formal labor force participation according to classic definitions offer valuable insights on women's entry into capitalist wage-labor markets. Then too, microdata analysis permits the researcher to take into account the entire household economy, such as the presence and work situation of a spouse, children, or other individuals, whether related or not. The integrated public use samples that we develop will enhance the original data by constructing composite household indicators of labor force participation. The hierarchical organization of the proposed census series, with individuals identified within household contexts, is well suited to the study of the household economy.

Analysis of the determinants of female labor force participation and child labor is a particularly compelling issue in Colombia, yet their study through time and across space is impossible with aggregate data. Integrated public use microdata series allow researchers to take control of the data, to move beyond the frustrations of changing technical definitions to focus on substantive issues of model development and hypothesis testing. Microdata are particularly salient where intellectual orientation and ideology ride roughshod over empirical evidence.

2. Demography of Violence. The Colombian death rate from violence—at some 80 per 100,000 population in the 1980s—ranks as one of the highest in the world for a country not at war (Pecaut 1997). One of the goals of the project is to standardize place-codes at the district level so that the demographic effects of violence may be studied at local as well as regional levels. While sampling error will be too great to study individual localities, with a uniform coding scheme it will be possible to develop appropriate aggregates. Integrated census microdata can be used to measure the effects of violence on types of communities, families, households, and individuals through time (Murillo-Castaño 1991; Ruíz and Rincón 1996). We have contracted with the Colombian statistical bureau (DANE) to ensure consistency in the coding of small places.

The integrated database includes variables on orphanhood, widowhood, and child mortality. The 1993 census was designed to measure a wide range of physical disabilities. Whether these questions will be continued in the 2001 census has not yet been decided. The Colombian census question on employment includes a special response for the disabled since 1993 and a decision has already been made to retain this option in the 2001 enumeration. The allegation that one-in-forty Colombians is a refugee may be tested with census microdata at both the national and local levels, but sustained research on this subject awaits resolution of inconsistent geographical codes for minor civil divisions.

3. Emigration and Immigration. In terms of international population movements, Colombia is primarily a country of emigration, with the United States constituting the principal destination country. Since 1985, Colombian censuses request information on mother's number of children resident abroad. Questions on retrospective migration also tap into movements to and from the United States. Coupling these data with the US-IPUMS databank will make it possible to compare Colombians resident in the United States with those resident in Colombia. The hierarchical structure of both datasets facilitates the study of individuals in their family and household contexts, so that it will be possible, for example, to study the correlates of unmarried Colombian fathers or mothers, whether they reside in the United States or Colombia.

4. Fertility. From the early 1960s to 1997, the Colombian total fertility rate declined from an average of 6.8 children to 3.0. This astonishingly rapid transition has elicited a great deal of scholarly attention (Puyana 1985; Flórez Nieto 1996), so much so that one might conclude that little work remains to be done—but this would be wrong. The Colombian PUMS, once suitably integrated, will permit the study of differential fertility patterns during a period of great fertility decline, and the relative impact of occupational class, region, education, size of locality, family structure, and a host of other variables at the individual, family or community level. The richness of these data will greatly enhance our ability to analyze the determinants of fertility decline in a developing country, and this may in turn lend insight into fertility control in late-developing regions. For women of child-bearing ages, Colombian censuses consistently report children ever-born, children currently alive, and the last born child's date of birth and survival status. In addition, the integrated microdata series will incorporate a set of fully compatible links between mothers and their children, and this will eliminate the most onerous aspect of one of the most widely used methods of fertility research, the own-child method.

5. Life Course Analysis. Changes in the timing of major life-course transitions—such as leaving school, leaving home, starting work, marrying, and establishing a separate household—have been studied in Colombia using retrospective survey data (Flórez Nieto 1989; Flórez Nieto and Hogan 1990; Flórez Nieto, Echeverri Perico, and Bonilla Castro 1990). While these studies have yielded valuable insights on the dramatic changes taking place in Colombia, our understanding of these processes would be vastly enriched by the analysis of all Colombian regions. Gutierrez de Pineda's path-breaking study of the Colombian family with her four typologies of the Andean, Black, Santanderan, and Antioqueñan cultural forms has never been put to the empirical test of nation-wide probability samples (Gutierrez de Pineda 1968). The integrated microdata series will provide the opportunity to test her models through a cohort analysis of the timing of change and differences by regions and among sub-populations, such as migrants and non-migrants, and more educated or less.

6. Aging and population projection. Demand for more sophisticated population projections increases as populations age (Banguero and Castellar 1993). A new multi-dimensional method for projecting populations and households requires input data that are readily derived from integrated public use samples (Vaupel, Yi and Zhenglian 1997). This summer, the principal investigator will attend the Max Planck Institute for Demographic Research workshop on family/household modeling and applications to evaluate this new method of projection and to ensure that the Colombian PUMS are properly designed to provide the necessary data.

These topics are intended only as representative examples of the sort of research that can be carried out with the Colombian integrated microdata series. Other key areas of investigation include the transformation of class structure, urbanization, internal migration, nuptiality, education, and the availability of public services such as water, electricity, sewage, and the like.

The large size of public use samples increases their versatility by permitting analysis of small population subgroups, metropolitan areas and other civil divisions. Policy analysts have traditionally focussed on short-run change, but there is increasing recognition of the need to distinguish long-term secular trends from temporary fluctuations. Public use samples will also allow Colombian policy analysts to set their investigations of department and local conditions in a comparative national context.

In summary, the Colombian census enumerations include a great deal of information on demography and socioeconomic structure that can only be exploited through analysis of microdata. At present only the broad outlines of the social transformations underway in Colombia are understood. Published sources provide limited information on topics such as fertility behavior, urbanization, household composition, and occupational structure. The proposed integrated microdata series for Colombia would allow the construction of comparable cross-tabulations on a wide range of topics that were not covered by census publications or were incompletely tabulated. Perhaps even more important is the potential for pooled multivariate analyses opened up by the availability of compatible microdata. Used in combination, the four data sets spanning a quarter century of cataclysmic social and economic change will comprise our most important resource for the study of the evolution of Colombian society. Violence has isolated Colombia from the international scholarly community almost as effectively as a blockade. This project will help circumvent that blockade.

Robert McCaa, the principal investigator, has been working on Latin American demography for almost thirty years. His first publication was an edited volume of census tables for the 1940 Chilean population census which had been published sporadically in the Chilean Statistical Bulletin over a period of years (McCaa 1972). Over the past fifteen years, he has recovered several large historical censuses, primarily for Mexico (McCaa 1984, 1988, 1989, 1991, 1997a), including a study of one of the most sophisticated censuses from ancient times, a sixteenth-century listing from Mexico written in Nahuatl (McCaa 1996).

Motivation for creating and integrating Colombian censuses comes from his experience—and frustration—over the past seven years as a user and a producer of large census microdata sets. An attempted comparison of ethnic intermarriage in New York City and Buenos Aires was blocked by an inability to gain access to the Argentine microdata. The data for New York City convinced him of the power of these data for addressing important historical issues and policy matters (McCaa 1993). Saving Latin American census tapes from destruction became an urgent concern; developing integrated public use samples, a goal. McCaa has already undertaken exploratory integration of portions of two Colombian censuses (McCaa 1998). He and his students have work in progress using Mexican and Chilean census samples (McCaa and Mills, in press).

Steven Ruggles also has extensive experience with census data. He has directed projects to create new public-use microdata samples for the United States census in 1850, 1860, 1870, 1880, 1900 and 1920, as well as an oversample of the Hispanic population in 1910 (NSF-SBR-9210903; NIH-R01 HD34572; R01 HD25839; R01 HD36451; R01 HD29015; R01-HD32325). In addition, he has directed a series of projects to create the Integrated Public Use Microdata Series (IPUMS), which incorporates census microdata from the U.S. Censuses for the period 1850 through 1990 (NSF- SES-9118299; SBR-9422805; SBR-9617820; NIH- R01 HD34714). He has published a variety of papers related to this work (e.g Ruggles 1991a, 1991b, 1991c, 1993, 1995a, 1995b; Ruggles, Hacker and Sobek 1995; Ruggles and Menard 1990, 1995; Ruggles and Sobek 1997; Ruggles, Sobek and Gardner 1996). He has also carried out extensive research using census microdata (e.g. Ruggles 1988, 1994a, 1994b, 1997).

As a result of these projects, we have become intimately familiar with the intricacies of public use samples, for both the United States and Latin America. In the course of constructing the US-IPUMS, our staff has invested thousands of hours in the reconciliation of variables such as relationship to head of household, occupation and birthplace.

Prof. Susan De Vos is one of the most accomplished researchers in the United States in the use of Latin American census microdata. She brings years of experience to the project of struggling with incompatible census definitions, both with Colombian census data, in particular, and Latin American censuses in general. As her letter of acceptance indicates she has a profound knowledge of the intricacies of changing Colombian census definitions and the careful work required to produce a well integrated databank of Colombian censuses.

Prof. Joseph Potter conducted the first technically sophisticated evaluation of enumeration error of a Colombian census, that of 1973 (Potter 1976). His expertise will be particularly valuable to this project, not only because of his work with Colombian data, but also for his wide ranging interests in fertility, mortality and health issues.

Prof. Carmen Elisa Florez is one of Colombia’s most distinguished living demographers, and certainly the most active researcher in the use of Colombian census microdata. In addition to consulting on the census of 1985 and working in various technical capacities at DANE, she has published a large number of technical, methodological and analytical studies using Colombian microdata. Her current project on occupational change in Colombia will provide valuable expertise in the difficult area of standardizing occupational codes.

Dirk Jaspers Faijer has almost twenty years experience at CELADE, having risen through the ranks from "associate expert" (San Jose, Costa Rica) to "chief of division for training on population" (Santiago, Chile). Lately, he has been responsible for the development of CELADE’s REDATAM initiative. Meanwhile, the CELADE tape library is under his direct supervision. He enthusiastically welcomes our project, and the opportunity to develop a mirrored web-site at CELADE.

The principal tasks required for creation of the Colombian Public Use Samples and integrating them into microdata series are given in Table 1, together with an estimate of the hours required for each task. This section describes each of these tasks in turn, with particular attention to design strategies to minimize problems of comparability.

1.0 Preparation and Design. Detailed planning of the integrated public use microdata series is a major undertaking, and design issues will be the focus of our work during the first year of the project. The following sections are intended to raise the most important design problems and our general approach to them.