Application of Latent Class Analysis to Classify Household Poverty in Development Studies: A Case Study Using Ghana’s 2021 Population and Housing Census Data

Traditional poverty assessments often rely on thresholds which can obscure the nuanced realities of multidimensional deprivation and the heterogeneous experiences of households. This study aims to empirically identify and characterise distinct multidimensional poverty profiles among Ghanaian households beyond widely used pre-defined cut-off points, and to determine the social, economic, and demographic characteristics that predict membership in these identified profiles. The household census data from Ghana’s 2021 Population and Housing Census was used. Latent Class Analysis (LCA) was employed to identify unobserved groups based on patterns of deprivation across various indicators (e.g., access to ICT, refrigerator, water, housing materials, toilet, cooking fuel, light, room density). Multinomial logistic regression was conducted to examine the influence of household characteristics on the likelihood of belonging to each identified latent class. The LCA model revealed three distinct multidimensional poverty profiles – relatively non-deprived" (34.72% of households, class 1), severely deprived (15.55%, class 2), and moderately deprived (49.73%, class 3). Male-headed, youth-headed, and aged-headed households, and rural, northern, and middle zone residence, were significantlly associated with higher risks of being in classes 2 and 3. This study demonstrates the utility of LCA in providing a nuanced, empirically grounded understanding of multidimensional poverty in Ghana.