Data Quality in Digital Small-Scale Fisheries Monitoring: Error Patterns, Training Effects, and Lessons from Zanzibar

Digital data collection systems have potential to enhance the effectiveness of SSF monitoring in low-data environments. However, transitions from manual to digital systems create new categories of errors that are not well-documented in existing literature. The study assesses the quality of collected data pre- and post-training sessions aimed at addressing enumerator data collection errors in SSF monitoring in Zanzibar conducted after the adoption of Peskas, an SSF monitoring system that utilizes KoBoToolbox for data collection. The system-generated validation logs obtained from a digital data portal were used to examine 10 error categories across 28 enumerators during a six-month period (February-July 2025) in 30 landing sites in Unguja and Pemba Islands. Analysis of 1,070 validation alerts collected prior to the training shows that length measurement errors under minimum species threshold are the most common category (44.5% of all errors), while individual fish count errors and missing quantitative data on identified taxa follow (21.6% and 11.5%, respectively).

A targeted feedback training to the enumerators led to a 77.8% reduction of total alerts down to 238. The greatest improvement in error reduction occurred in the case of under- or overestimation of bucket weights (97.6%), individual counts (94.4%), and missing quantitative data (83.7%). Contradictorily, the errors of length-above-maximum increased after the training period; thus, there is evidence of higher levels of measurement involvement rather than regression in errors. Thematic analysis of the qualitative feedback of enumerators resulted in the identification of four major cross-cutting areas of challenges: lack of equipment, species misidentification, low level of awareness of fishermen, and access problems. The study proves the efficiency of an iterative feedback-based capacity development approach based on systematic error data for improving digital fisheries data quality.