A Confounded Statistics

In 1898, the Mexican Agricultural Society proposed a comprehensive survey of Mexico's agricultural properties to be presented at the Paris Exposition in 1900. In collaboration with the national Department of Fomento, or Development, they composed a table and circulated it to all 2,300 municipal governments in the country. Across 1899, about 1,400 municipalities returned their data. Blaming recalcitrant local governments for failing to send back the requested information, the Agricultural Society and Fomento set the project aside and stashed the surveys that had been returned in the archives.

Over the past years, research assistants and I have transcribed, cleaned, and digitized the surveys that made it into the Mexican national archive. The information they included makes clear that it was municipal officials' divergent means of responding to the tables, not just their failure to complete them, that kept the project from completion. Eagerness to take part as well as recalcitrance confounded this project of state knowledge making.

The formulation of the table and requested information seems easy enough to parse.

Sample table returned by municipal officials from Burgos, Tamaulipas. AGN, Fomento: Exposiciones Exteriores, caja 51, exp. 8.

In total, the table represent about 14,000 properties spread across 18 states.

Eventually, I will make this data available via the  ArchivoMex  project hosted at Mexico’s Laboratorio Nacional de Políticas Públicas and the Inter-university Consortium for Political and Social Research (ICPSR). For now, feel free to navigate some of the maps below to see data summarized at the municipal level as well as a sample of the property-level data from Coahuila and Michoacán.

As my article "A Confounded Statistics" makes clear, scholars need to take caution in using this data. Despite cleaning and standardizing as much as possible, much of the information remains incommensurable. Even within states, municipal officials filled out the survey based on their own understandings of what qualified as a property, how to report different kinds of crops and yield, what information was being requested about machinery. The data, summarized to the municipal level or accessed property by property, is valuable and provides an overview of rural Mexico in 1899 that we have not had access to before, but it is neither as comprehensive or as comparable as the maps below might suggest. There are good reasons it was never published.

Maps from the 19th century do not have a great deal of precision when it comes to municipal boundaries as few had been surveyed and definitely marked. To allow for visualizing the data as polygons (that is, shapes) rather than dots, we have matched historical municipalities to contemporary ones, using shapefiles produced by the  Instituto Nacional de Estadística y Geografía (INEGI)  in 2020. Municipal names have changed in the century since the data was produced, but we have matched all but 60 of the municipalities in the dataset to their contemporary equivalents. Also, some municipalities have expanded while others have divided, meaning the shapes are not exact representations of historical municipalities and some shapes represent more than one municipality. Over time, we hope to increase the precision of this matching and publish a set of GIS layers that allow scholars to more easily map their nineteenth century data.

We are also producing a set of shapefiles representing 19th-century administrative districts eliminated in the wake of the Mexican Revolution by georeferencing and tracing maps from 1899 and 1904. The map above was produced by the students in my Spring 2021 Making Maps of Mexico course at Johns Hopkins, and Martin Salmón of CIDE/Archivomex is completing the work for the rest of the country.

ArcGIS visualizations based on summary data at the municipal level allow us to see patterns in things like the average size and value of properties - larger, more valuable properties in the north, smaller, less valuable properties in the south. Oaxaca here stands out as including a number of large but low value properties - clicking on some of these circles we see that many represent villages where the entirety of the town's land was reported as a single property. Clicking on a circle also reveals information about total and average size, number of properties, irrigation, and forested land.

Here, I have used ArcGIS to highlight the predominant climate descriptor by municipality and mapped that on top of a modern map of North America's climate zones. Temperate clearly meant something different in Chihuahua than it did in the central plateau, and visualizations allow us to see how local relational context mattered in what municipal officials reported. Click on the municipality and you will see charts representing the climates, kinds of property, and crops reported.

This map summarizes average wage data for men at the municipal level, visualizing the range of wages across the country and the lack of concentration of high wages in any particular region.

This dashboard provides summary data as well as individual pop ups for reach property listed in the states of Coahuila and Michoacán. Click on the name of a municipality to zoom in, and then click on the dot within that municipality to see the properties reported there. The arrows at the top of the pop up will move you through each property in turn.

Because we do not know the locations of the properties within each municipality, for now they are stacked on top of each other. I am considering dispersing the datapoints across each municipality, but worry this will give a false impression of having located each in space. Also, as a quick note, not all dots sit within the proper municipal shapefile as the matching is still a work in progress.