Evaluation of a participatory approach to stratifying neighbourhoods across the Tanzanian city of Dar es Salaam in terms of income by comparing with latent profile analysis of deprivation based on national census data.

Studying geographic heterogeneities in the characteristics of city neighbourhoods, such as population income and deprivation, has been a common practice in urban health studies for tailored and targeted interventions, especially in highly developed and diversified cities. While a range of different approaches has been applied to classify different parts of towns and cities, participatory stratification approaches (PSAs) have become popular despite their subjective basis because of their affordability, simplicity and practicality, all of which allow them to be frequently updated. While more objective statistical approaches, such as latent profile analysis (LPA), can also be used to stratify neighbourhoods using formal socio-economic and demographic data, these rely on the availability of rich datasets and advanced analytical capacities that are not always available in low and middle-income countries. This study assessed a PSA to stratify neighbourhoods across the Tanzania city of Dar es Salaam in terms of income, by comparing it with a complementary LPA using national census data from 2012 to stratify them in terms of deprivation. A consultative community-based workshop was used for the PSA, while 15 selected deprivation indicators from the census data were used to profile them using LPA. While the PSA allocated neighbourhoods to five income strata, six clear deprivation strata could be distinguished by LPA. A strong positive correlation was observed between the stratum identified by the LPA and that obtained through the PSA ({rho} = 0.739, p < 0.0001). Furthermore, paired comparison of the two sets of correlation coefficients between each deprivation indicator and the stratum assigned by each stratification approach revealed no difference (V = 33, p-value = 0.1354), confirming that the two approaches yielded very similar patterns of stratification., Also, the two approaches yielded broadly comparable cartographic pictures of the city, depicting similar spatial distribution of wealth and poverty. Overall, this evidence indicates that subjective community knowledge and lived experience may be invaluable for understanding the built environment and for mapping out pockets of poverty and affluence at fine scales with minimum resources.