Exploring diversity - Visualizing the Afrozensus Germany in R
CFE’s division Advocacy for Inclusion focuses on changing structures, processes and policies to combat discrimination and foster diversity and inclusion. In order to achieve this, CFE researchers have developed an intersectional survey tool that captures the entire spectrum of the German Equal Treatment Act (Allg. Gleichbehandlungsgesetz). It aims to collect detailed data on diversity and discrimination in private and public organizations as well as in specific societal sectors.
With this tool, CFE wants to support, among others, organizations, businesses, the public administration in working more inclusively, and, ultimately, in embodying our society’s full spectrum of diversity. CFE’s research enables decision makers to understand the diversity of their personnel as well as the occurences of racism that are endemic in their organizations. It also provides specific tools to address discrimination. In this way, the tool enables the first step for a fundamental change towards more diverse structures and people.
Together with EOTO e.V., one of Germany’s leading Black empowerment organizations, Citizens for Europe has expanded this tool to be used with Germany’s Black population through the Afrozensus – CFE’s largest project to date. The aim of the Afrozensus is to obtain as comprehensive a picture as possible of the experiences of people of African descent in Germany, how they assess their lives in Germany and what they expect from politics and society. The findings of the survey were to be made available to relevant communities and policy makers.
In order to analyze and visualise the Afrozensus data, CorrelAid supported a workflow creation in R that allows users to analyze different data sets and question types in various forms. As many employees in the organization were not trained in using programming approaches, usability was important. Hence, the goal of the project was to develop an easy-to-use, well documented R package that would facilitate the individual analysis steps. By doing so, the organization hoped to speed up and standardize data exploration while allowing the analysis to be reproducible and documented.
- Package = set of pre-defined functionalities in specified coding languages
CorrelAid volunteers worked with a synthetic dataset that mirrored the actual Afrozensus data with its over 6000 responses and 700 columns. By their nature, the variables represented different types of survey questions: single-choice, multiple-choice, matrix, numeric, and open-ended response questions covering topics such the eight dimensions of the German Equal Treatment Act, experiences of discrimination in different areas of life, volunteering and activism.
The CorrelAid team has successfully developed an R package containing various ggplot2 visualization functions tailored specifically for Afrocensus survey results. With the package, researchers only have to specify a minimal set of input parameters, e.g. which question to analyze. The package will then create pre-defined visualiaztions, automatically adapting the type of visualization to the type of the question, e.g. a histogram for numeric questions or a bar chart for single-choice questions. Visualizations can be created individually or in bulk in an automated fashion. The visualizations take into account the style of the Afrozensus final report and, therefore, can also be used for the final report.
For the report, the CorrelAid team created an initial examplary R Markdown exploiting the package functions to build the visualizations automatically for multiple survey variables, preparing all the necessary material for the data exploration. The created R package was tested by the team manually as well as with automated unit tests.
The package was handed over in a meeting where the CorrelAid team supported the Afrozensus researchers with information on how to use the package and its functionality.
The package allows the researchers to focus on what matters most – searching for insights and patterns in the extensive survey responses and identifying those aspects that make life in Germany more challenging and how these aspects interact with each other by looking at and comparing various visualizations.
The solution provided by the CorrelAid team requires minimal technical knowledge, yet allows for customization as needed. Thanks to this focus on usability and ease of use, interactive data exploration and visualization is now also possible for researchers with less technical skills. And researchers with a more technical oriented skillset can now iterate even faster during data exploration.
Besides enabling interactive data exploration, the package has also facilitated more automated approaches. Over 200 reports on individual variables in the Afrozensus have been automatically generated using the package.
Thanks to the data science team at Citizens for Europe, the package will continue be actively maintained and developed in the future – beyond the scope of the Afrozensus project.