MEASURING GEOSPATIALLY EXPLICIT PERCEPTIONS OF SUDANESE ETHNIC GROUP LOCATIONS : A COMPARISON OF SUBJECT-MATTER-EXPERTS AND ONLINE DATA

The geospatial perceptions of Sudan and South Sudan subject-matter-experts (SMEs) regarding the location of Sudanese ethnic groups were collected and analyzed, in order to determine how to synthesize and gain meaning from multiple geospatially explicit responses. This study utilizes and attempts to build on methodologies based in the geographic sub-fields of participatory mapping and participatory geographic information systems, as well as, anthropology. To determine how well the previously examined SME perceptions of ethnic group location compare to online news articles, methods based in network analysis, were applied in an attempt to verify the SME perceptions.   Geospatially explicit response maps were collected at the Sudan Studies Conference (April 2012) depicting SMEs' perceptions of ethnic group location. From the hand drawn response maps, digital raster layers were created, aggregated and displayed using graduated colors. The Geospatial Similarity Analysis found that areas where seven or more SMEs agreed an ethnic group was located, was the minimum amount of overlap necessary to indicate agreement. A geospatial consensus analysis was applied using the cultural consensus model (CCM) to determine if the respondents' percentage of overlap with one another indicated culturally shared knowledge, concerning the location of ethnic groups. A geospatial cultural consensus or shared knowledge was found for two of the four ethnic groups analyzed.   To verify the locations indicated by SMEs, an independent data source was used in a geospatially linked semi-automated network text analysis, paired with a content analysis. These analyses were used to identify co-occurrences of location names and ethnic group names within articles from the Sudan Tribune. The Euclidian distance between expert-indicated location and locations cited in the Sudan Tribune were determined in order to characterize their agreement geospatially. To gain a more comprehensive understanding of the co-occurrences of ethnic groups and locations, each one was examined and contextually classified. The co-occurrences contextually classified as indigenous lands proved to have greater geospatial agreement with the location indicated by SMEs, more so than any other contextual classification (political conflict, ethnic conflict, resources, oil and history).   Measuring consensus geospatially among multiple informants can be applied to any type of geospatial knowledge, in this case the representation of expert perceptions, making it a valuable addition to participatory mapping methods. Semi-automated network text analysis that codes for contextually indicative terms could be further developed and used to examine and model the distribution of ethnic groups in a more comprehensive manner.  M.A