Data Collection

The first step of the project was to collect and align several large-scale, unpublished datasets (Table 1). We further drew on secondary sources to improve the geographical coverage of the database. In some cases, we assembled data directly from lists and tables in the paper/book in question, whereas in other cases, we carefully transcribed information on the events mentioned within the pages. In addition, we occasionally undertook detailed searches for social conflict in archival sources.

Table 1.

Overview of the Main Existing Datasets Included in HiSCoD

Note: Bohstedt (Reference Bohstedt2010) provides lists of food riots for various years between 1347 and 1819. For more information on the content of each dataset, see Section B of the Supplementary Material.

Events recorded in the HiSCoD database come from a wide range of primary documents. For instance, information on medieval and early modern social conflict is found in chronicles, annals, journals, judicial records, and diaries, whereas information from the eighteenth and nineteenth centuries are retrieved from administrative correspondence, judicial records, municipal council records, private papers, and/or local newspapers.Footnote ⁶ Therefore, the level of information available on individual instances of social conflict is heterogeneous and disparate. It is sometimes possible to identify the composition of the crowd and to follow the course of events hour by hour through police reports and trial records, while in other cases evidence identifies only the location of the event. To address this imbalance, we use a simple form to standardize existing databases and collect new data. We also chose to rely on a broad definition to qualify an event as a social conflict in order to grasp the infinitesimal episodes of discontent and include most of the works previously carried out by historians.

Definition

The issue of what comprises a social conflict is one of the most highly debated topics among historians who used different metrics and definitions to establish their datasets. It is difficult to reach a consensus on the characteristics and criteria required to define an event as a social conflict because they differ vastly in form and scale. In his study on food riots in England, Bohstedt (Reference Bohstedt2010, 17) doubted that “anyone will ever compile a complete record of all the riots in a period” because it is “impossible ‘by definition’ for two lists to agree if they did not agree on a definition of the events they were collecting in the first place.” This was precisely the task we set ourselves: to merge and combine these disparate datasets into a unified, analysis-friendly format within a database.Footnote ⁷

We used the expression “social conflict” as a generic category to encompass all the expressions used by scholars and/or mentioned in the sources. We followed Nicolas (Reference Nicolas2002, 75) and considered social conflict to be any event involving a group of at least three individuals belonging to different families and which either perpetuates violence or threatens violence against one or more members of a different group or against representatives of political, religious, and economic power; or any event involving an attack on property, buildings, furniture, papers, or other signs symbolizing such powers. We chose to use a low threshold on the number of participants not only to grasp infinitesimal episodes of discontent, but also to allow for the inclusion of other datasets that used different thresholds. For instance, Bohstedt (Reference Bohstedt2010) only considers events with more than 50 participants, whereas Bercé (Reference Bercé1974) listed events with more than 20 participants.Footnote ⁸ In Section B of the Supplementary Material, we report the definition and threshold used by scholars to assemble their database in order to guide the user and ease the assessment of comparability of data across time and space.

Typology

The main objective of the HiSCoD project is to act as gateway for the study of historical social conflict by gathering existing information from many studies and sources. Pre-industrial popular disturbances included a broad spectrum of events from grain and bread riots to crafts revolts, tax riots, religious conflicts, and many more. They also involved a variety of participants, often with a range of different motives, with riots against high crop prices occasionally leading to further protests about wages or taxation, for instance, and/or conflict with local authorities about setting arrested rioters free. Tilly (Reference Tilly1976, 375) insists that “to categorize is a first step on the way to identifying what there is to explain, and therefore on the way to explaining it.” Every historian who has conducted a large-scale survey of historical social conflict has thus been confronted with the challenge of classifying the diverse events she has identified. This means that the literature includes dozens of typologies, sometimes involving very different categories depending on the social and historical context.Footnote ⁹

We addressed this issue by creating a simplified classification based on 10 categories to encompass a wide range of geographic areas and a long-time span (Table 2). We assigned each event to one category based on the information available and, when relevant, on the researchers’ original typology—which is also included in the database. We would highlight here that our classification is first and foremost intended as a tool to guide users through the myriad of events reported. One can used the various information included in the database to establish her own typology and/or assign several categories to an event.

Table 2.

Classification Used in the HiSCoD Database

Note: We categorized events related to seigneurial taxes as tax riots rather than feudal conflict. In 333 instances, the lack of information in the sources pre-empted the attribution to an exact category.

Structure of the Database

Whenever possible, we created one entry for any social conflict on a specific day in a given location.Footnote ¹⁰ To situate each event in space and time, we recorded time units (year, month, and day), historical and contemporary location, and geographic coordinates (longitude and latitude), along with an indicator for geographic precision.Footnote ¹¹ In addition to the summary of the course of events, we added two variables: the number of participants and an indicator for the participation of women. Although some scholars have reported evidence on the characteristics of participants, such as age, occupation, and social status, this information is often incomplete and could suffer from reporting biases. Therefore, we chose not to create such variables and report existing information on participants in the description of the event. We made this choice to keep the structure of the database user-friendly. For all events from an existing dataset, we included the unique identifier of the original database to allow easy matching of HiSCoD entries with the original ones. Table 3 provides an overview of information available for each entry.Footnote ¹²

Table 3.

Overview of Information Available in the HiSCoD Database

Data Preservation

We designed the HiSCoD project with the aim of creating a platform to ensure that data on historical social conflict were preserved and to make sure that such data can be easily located, accessed, and reused. We chose to encode the database using the EAD standard. The Encoded Archival Description standard is a public domain XML schema created and used by archivists for encoding archival finding aids. Its markup structure makes it a machine-readable format, enabling data to be shared easily across platforms. This solution guarantees interoperability and that data will be preserved and accessible.

Data Accessibility

Users can access data in one of three ways. First, they can navigate the interactive map on the project website to identify events of interest and access individual forms, which can be manually downloaded as pdf files in both English and French.Footnote ¹³ Second, they can access a user-friendly semantic repository, based on the EAD standard, to either search for keywords or explore data by interacting with a data access system. Finally, the most up-to-date version of the dataset can be download as a csv file from the project’s GitHub repository.Footnote ¹⁴ A detailed description of the variables included in the dataset is available in Table A.1 in the Supplementary Material.

Data Contribution

In Section D of the Supplementary Material, we provide detailed explanations on how to contribute to the HiSCoD project.

Data Overview

Figure 1 shows the distribution of historical social conflict across countries from the European High Middle Ages to the second half of the nineteenth century (c. 1000–c. 1870). As of June 2023, most events are located in Europe, and in particular, France and England (about 92%).Footnote ¹⁵ This bias is due to the fact that we first gathered events from existing databases that were not previously published or easily accessible (e.g., Bohstedt Reference Bohstedt2010; Lignereux Reference Lignereux2008; Nicolas Reference Nicolas2002). The absence of recordings of social conflict for a country should not, however, necessarily be interpreted as resulting from missing data. It might also be the result of the historical context particular to a country. For instance, whereas food riots were endemic in France and England during the eighteenth century, they were relatively uncommon in Spain before 1766 (Rodríguez Reference Rodríguez1973, 145).

Figure 1.

Geographical Distribution of Social Conflict in HiSCoD

Note: Only countries with at least 25 entries are displayed.

Figure 2 shows the temporal distribution of the events recorded in the database. The unequal distribution of social conflict originates both from survival bias in relation to sources and available data, and from the fact that the seventeenth and eighteenth centuries witnessed evolution and transformation in the forms of popular disturbance (Tilly Reference Tilly1976). Overall, users should be cautious before using data for comparisons across time and space. While it is possible to run quantitative analysis on various subsets of data, the HiSCoD project is first and foremost conceived as a gateway to preserve and access information on historical social conflict, and does not provide a perfectly comparable worldwide dataset. Each dataset has its specific characteristics and its own bias as they are all based on the definition and categorization established by individual scholars who collected and classified events (see Section B of the Supplementary Material).

Figure 2.

Number of Social Conflicts per Quarter Century

Figure 3 provides an overview of the number of entries for which we were able to document the main variables included in the database: number of participants, participation of women, summary of the course of events, archival signatures, and bibliographic references. More than twelve thousand entries (58%) contain a description of more than five words in length in either English or another language. Unfortunately, in most cases, it is impossible to document the number of participants and/or the participation of women with any certitude due to the paucity of information in the sources.

Figure 3.

Available Information for Individual Social Conflicts

Note: The database includes 9,616 entries for which we can ascertain the involvement of women. The database includes 12,203 entries with a description of more than five words in length. The database includes 2,935 entries with at least two bibliographic references.

Data Use Cases

A number of scholars have already used historical data on social conflict to study a wide range of questions in economics, history, and/or political science. For instance, Aidt and Franck (Reference Aidt and Franck2015) investigate the effect of exposure to violent social conflict on voting behavior during the 1830s in England. They use data on the Swing riots to show that voters’ and patrons’ fear of rioters in the immediate neighborhood of their constituency shifted their allegiance toward the reform-friendly Whigs during the 1831 election. In another paper, Aidt, Leon-Ablan, and Satchell (Reference Aidt, Leon-Ablan and Satchell2021) use the same dataset to analyze information flows and the impact of news about repression on participation in riotous events. They also document the role of local organizers and personal and trade networks in the geographic distribution of these riots.

Although replication is now a relatively common practice in social sciences, replication materials rarely include raw data, which makes it difficult to reuse them beyond the scope of the publication. We hope that this ongoing project will facilitate new research on social conflict during the pre-industrial era by improving access to original data.