Print Friendly vers. Spanish version

History of the Guadalajara Censuses Project 

The roots of the GCP go back to 1978. While researching in the Guadalajara Municipal Archives, the project director, Rod Anderson, became aware of the existence of a considerable number of city population censuses (called padrones), including one nearly complete for 1821 and a less than complete, but rare followup census for 1822.¹ Although various scholars knew of the existence of the data, only Berkeley historical demographers S. F. Cook and Woodrow Borah had published from the material, using partial, hand-calculated returns.²

In 1983 Anderson constructed a machine-readable database from a systematic sample of every tenth household taken from selected wards (cuarteles) of the two censuses, from which he published a monograph and a number of articles.³ In 1990 the PD and several of his graduate students began preliminary work on the current project. They were Elaine Carey, Gerry Geis, Burton Kirkwood, David Sicko and James (Jay) Tapp. In order to provide an institutional framework for this and other projects, in 1992 Anderson and his graduate students created the Urban History Workshop (UHW). The Workshop focuses on providing students practical experience in research design, methodology, scholarly writing and publication. In keeping with its emphasis on hands-on experience, the UHW founded the Urban History Workshop Review in the fall of 1993. When the editor of the UHWR moved on to the History Department of the University of Evansville, the latter co-published the journal along with Florida State.

In order to test our project format and procedures, the UHW created a preliminary database utilizing EXCEL spread sheet software to create a fixed structure rectangular data file. By the summer of 1993 data entry had been accomplished for 4147 individuals (758 households) from three of the city's twenty-four 1821 districts or approximately 11 percent of the city's 1821 population (38,091). Although it was not a representative sample, the preliminary database led to a number of codebook design innovations and enabled the staff to gain experience in coding several constructed variables, the most important of which were the household and family structures. In general the preliminary database was instrumental in modifying, and in some cases substantially change, our working procedures, coding techniques and software requirements.⁴

After the completion of the preliminary study, in the fall of 1993 and the following spring the UHW organized a series of three, three-day workshops to critique the proposed project, bring in four outside scholars from different fields and expertise: Margo Anderson, Department of History, U. of Wisconsin-Milwaukee; Michael Hawthorne, Department of Political Science, Pembroke State U., NC; Robert McCaa, Department of History, U. of Minnesota; Michael Scardaville, Department of History, U. of South Carolina. Anderson is a leading scholar of the U.S. historical censuses. Michael Howthorne, a political scientist, is an expert in research design. A Latin Americanist, Robert McCaa, is one of the field’s leading historical demographers and brought a wealth of practical experience and theoretical knowledge to our project. Michael Scardaville is a Mexicanist with long experience in urban history whose experience with the data analysis software SAS complemented Hawthorne and McCaa’s familiarity with SPSS, the latter probably the leading data analysis system used by historians.

In addition also participating were: Charles Nam, nationally-known demographer from FSU’s Center for Population Studies; C. Peter Ripley, Department of History and Director of the NEH-funded Black Abolitionists’s Papers Project; Morton Winsberg, Department of Geography and expert in spatial aspects of contemporary census data; Douglas Charity, technical advisor to the project from the beginning. Although disagreeing on various details, the workshop participants made a number of specific suggestions.⁵

Soon afterwards, the UHW formally constituted the Guadalajara Censuses Project and assembled an international Board of Advisors. In particular, the project wanted to enlist a wide range of views from various disciplines and from non-quantifiers as well as “number-crunchers.” To that end, historical geographer, David Robinson and French historical demographer, Thomas Calvo, agreed to serve on the Board, as did Luís González, perhaps Mexico’s foremost living historian and very much self-consciously not a social scientist. Other members are Eric Van Young, well known colonial historian and also not a quantifier; Asunción Lavrin, senior scholar on gender; Guillermo de la Peña, Mexican social anthropologist and specialist on the economic history of western Mexico; and Carmen Castañeda, the leading historian of Guadalajara and the GCP’s coordinator in Guadalajara. Michael Scardaville provided his knowledge of the Mexico City census of 1811 and Robert McCaa his demographic expertise. Also agreeing to serve as local Board members were Ripley, Nam and Winsberg. The latter in particular has provided extensive aid and advice concerning our Geographic Informational Systems (GIS) component. Douglas Charity is the GCP’s long-time technical advisor and invaluable consultant. It has been an active Board, and its individual members have provided considerable advice and counsel.

The long term goal of the Guadalajara Censuses Project is to create a database from the population censuses of Guadalajara, from the military census of 1791 to the modern city census of 1930. For the purpose of teaching the GCP emphasizes graduate student involvement in all phases of the project: research design, methodology, paleography, data collection, coding, data entry, data verification, the writing of user guidelines, Website maintenance and the development of the project CD-ROM.

UHW staff began actively publicizing the Guadalajara Censuses Project in 1993, both through its annual bulletin, the Urban History Workshop Review, and through the Conference on Latin American history. In the Fall of 1993 and again, in the summer of 1997, the Project Director and the UHWR editor, J. Burton Kirkwood, spent several weeks in Guadalajara working on various aspects of the project. These activities and the GCP in general have been strongly supported by financial and other aid from the Department of History and the College of Arts and Science, of Florida State University, and from the F.S.U. Council on Research and Creativity. History chairs Neil Betten, Richard Greaves and Neil Jumonville have been particularly supportive and encouraging, as have Deans Larry Abel and Donald Foss of the College of Arts and Sciences.

The current phase of the project began in April of 1999. The then Director of the Division of Preservation and Access of the National Endowment for the Humanities (NEH), Helen Agüera, notified Rod Anderson that the GCP application for funding had been approved, requiring only minor modifications. The NEH funded portion amounted to $114,959 out of a total budget of $207,435, the remainder being provided as cost-sharing by the Department of History, the College of Arts and Sciences and F.S.U.’s Office of the Vice President for Research. The project consisted of creating a database from the padrones of 1821 and 1822, and ran from June 1999 to April 2001.

By the end of the project, the GCP had created a database of 89 variables for 57,091 cases, mainly individuals (the exceptions are several hundred vacant houses, which we felt useful enough to report), including many constructed variables created for the convenience of the users. These are the variables which constitute the database for the CD-ROM. We actually entered more data than we promised because in the course of the project it was decided to include data for the heads of households for six districts in 1822 for which no complete manuscript census had been found. We also entered separate lists of clerics living in religious houses, of families living in the parish charity house, and a list of all incarcerated persons. The total pieces of social data in the database are just over three million. Coding and data entry efficiency during both the Literal and the Constructed phases exceeded expectations, although the project required more time than we had originally planned. In part, the reason the project took longer than expected was that we added about twenty constructed variables to our original list of variables, several of which were relatively complex and all of which were labor intensive. We reasoned simply that given the great time and expense already undertaken, that we might as well make the most of it. Also, a good deal more time than anticipated was spent in data verification and error detection.

Data Verification and Consolidation. For a detailed discussion of the GCP procedures for data verification, please go to Error Detection and Verification Procedures. The search for data entry error was carried out at regular intervals during data entry, and at the end of the entry of each cuartel. During the literal phase a print-out of the entered data is compared to an enlarged, 11" x 17" copy of the original manuscript page. Numeric values were compared orally by two staff members. Literal data written in script was sight reviewed by one staff member. When all the data for each cuartel have been entered, the data was “cleaned.” This process consisted of creating frequencies for each variable which were then reviewed for anomalies in values. Tables were created for special variables prone to error. (In our first project, for example, this process revealed fifty-three male doncellas–“ young ladies.”) All given and surnames names were examined by a native speaker of Spanish to verify their likely authenticity. Once Literal data entry was completed for each cuartel separately, they were merged into one large file and underwent a further “cleaning” process in a final search for error.

In order to quantify the rate of error in the finished database and to identify those areas most vulnerable to error, we generated a random sample of 3412 individuals from our 21 literal entry variables (6% of 57091). The sample data was then compared to the original manuscript, case by case. Errors were noted (and corrected). To provide some comparison, we compared our rates of error to the error rates for the IPUMS U.S. national historical census project, the leading census project in the field.⁶ In the eight comparison variables for two censuses our error rates were lower than IPUMS, and in seven they were higher. On the average our error rate was 0.53 per 100 cases (mean) and 0.46 (median), or approximately half an error per 100 cases (a case being the data of one variable for one individual). After the completion of the constructed variable stage of our project, we took a sample of 3400 cases for each of three sets of variables (household and family; “numbers” and kin; and “locs” and migration), comparing the data entered to the coding sheets. Our average error rate per 100 cases for the eleven household and family variables in this category was 0.57 (mean) and 0.53 (median), with a low of 0.44 and a high of 0.82. Despite our efforts, errors will remain. We hope that all users will notify us of any errors they might find, so that they can be corrected in future versions of our database. (We would like to remind our users that the Archive File will contain errors since caught and changed in the Consolidated File.)

Current Project: Guadalajara Censuses Project

The GCP’s current project is to add to the 1821-22 database population censuses from 1791, 1811, 1813-14, 1824, 1838-42, 1850 and 1930. Unfortunately, unlike the magnificent census of 1821, those years suffer from certain problems. The “military” census of 1791 is rich in detail for those individuals surveyed, but do not cover the city’s extensive Indian and Mulatto population, since they did not have the “privilege” of serving in the local militia. Also, the returns for the years from 1811 through 1850 are missing for many of the city’s districts. In the case of 1930, the entire manuscript census is available but, because by 1930 the city’s population had reached nearly 200,000, the GCP can afford to enter only a 10% sample rather than a complete census.

Nonetheless, data from the earlier period in the history of Mexico is rare and worth analyzing even in partial form. Bracketing the five critical decades surrounding Mexico’s independence from Spain, this data will enhance the 1821-22 censuses by providing breadth of coverage, documenting urban life from the height of Bourbon rule to the chaotic years of the regime of Santa Anna. The 1821-22 censuses will serve to anchor the new data, as a base for comparison. And although the new censuses are incomplete, every one contains the manuscript census for Cuartel 8, the city’s most populous district and the heart of working-class Guadalajara (as it has remained down to the present). The sample for 1930 provides the rare opportunity to enable the database to touch three centuries. When fully integrated, the completed Guadalajara Census Project will comprise more than 130,000 cases, one of the largest, easily accessible historical urban census databases anywhere.

Also planned to be included in the final product will be a documentation file with the following six guides: (1) Guide to the GCP database, written by P.I. Rod Anderson. (2) Guide to SPSS and the GCP Database, by Margo Anderson. (3) Guide to Excel and other formats in the GCP database, by Douglas Charity (project technical advisor). (4) Guide to Genealogical Explorations of the GCP Database, by Rod Anderson. (5) Guide to Census Data, by Margo Anderson.

The current project is funded through a renewal grant of $166,838 from the National Endowment for the Humanities, Division of Preservation and Access, with a cost sharing of $290,651 from the Department of History and the College of Arts and Sciences of Florida State University, and the Office of the Vice President for Research. The project is projected to run from June 2002 through December 2004.

Notes:

1. The 1821 and 1822 original manuscripts are found in the Archivo Municipal Histórico de Guadalajara (AMHG). Of the city’s 24 cuarteles (districts) in 1821, data is available for 23. Only cuartel 16 is lacking, and that is available for 1822. Also located in the AMHG are partial returns for 1811, 1813-14, 1824, 1838-42, and 1850. A partial but extensive manuscript census from 1888 is found in the Archivo Histórico del Estado de Jalisco (AHEJ). A complete manuscript census for Guadalajara for 1930 is located in the Archvo General de la Nación (AGN), and has been microfilmed by the Family History Library, Salt Lake City. Also located in the AGN is a manuscript census for 1791 called the “military” census since it includes data only for those individuals and families who were eligible for the royal militia. Indians and Mulattoes were not eligible. All have been copied by the GCP. A copy of the 1791 “military” census was provided to the GCP by Prof. Eric Van Young, University of California at San Diego. In addition the GCP has a copy of a business census for Guadalajara in 1880 and an industrial census for the city in 1907, the original documents for the latter two being located in the AHEJ.

2. Most notably Sherburne F. Cook and Woodrow Borah, Essays in Population History: Mexico and the Caribbean, vol. 1 (Berkeley, CA, 1971-74). See also Luís Paez Brotchie, Guadalajara, Jalisco, Mexico. Su crecimiento, división y nomenclatura durante la época colonial 1542'1821 (Guadalajara, 1951).

3. The monograph is Guadalajara a la consumación de la Independencia: estudio de su población según los padrones de 1821-1822 (Guadalajara, 1983). See also “Race and Social Stratification: A Comparison of Working-Class Spaniards, Indians, and Castas in Guadalajara, Mexico in 1821,” Hispanic American Historical Review, 68 (May 1988), 209-43.

4. Most critical was the switch from the EXCEL spread sheet to an SPSS Data Entry Module, and the decision to create a literal (verbatim) database, from which we would create the standard coded files for data analysis.

5.For a summary of workshop deliberations, see Urban History Workshop Review, no. 2 (1994).

6. W. Block and D. L. Star, “Data Entry and Verification,” Historical Methods, vol. 28:1. The variables compared were: Household number, Household size, race, sex, age, marital status, occupation, birthplace. The variables added to those to obtain our averages were: First name, surname, title and block number. The actual error rate was lower because we corrected the errors in the sample.

Search WWW Search www.fsu.edu