Inter-observer reliability in transect-based observations of environmental waste in Greater Accra and Kisumu: implications for waste management

Transect surveys are widely used to quantify mismanaged plastic waste discarded to the environment. However, very fewstudies have quantified inter-observer reliability in environmental waste surveys. The aim of this study was to assess inter-observer reliability for a set of environmental waste indicators derived from transect surveys in slum areas of two sub-SaharanAfrican cities. Pairs of observers independently recorded counts of scattered waste items, large waste piles, and burnt wastealong 64 environmental transects in Kisumu, Kenya, and Greater Accra, Ghana. Concordance correlation coefficients wereused to measure inter-observer reliability for derived indicators measuring mismanaged waste density, waste composi-tion, waste origins of policy concern, and waste disposal practices. Evidence across all observers consistently showed thatsingle-use disposable diapers, discarded Personal Protective Equipment, and bagged or bottled water are locally importantconstituents of mismanaged waste in both cities that should be addressed through urban waste management strategies. Inboth cities, there was generally excellent inter-observer reliability for density of burnt waste and density of large waste piles(concordance correlation coefficient > 0.9 for all but one observer pair), but weak to moderate inter-observer reliability forscattered waste density and densities of specific waste items such as disposable nappies and discarded Personal ProtectiveEquipment. Our study shows that beach litter survey protocols can be adapted for use in slums. However, to generate robustestimates of scattered waste, its composition, and waste origins of policy concern in slums, assessment of inter-observerreliability should be incorporated into field team training.