Poverty and maternal nutritional status as determinants of weight at birth: a multilevel ordinal logistic regression approach

Samson B Adebayo, Society for family Health
Oladipupo Banji Ipadeola, Society for Family Health
Jennifer Anyanti, Society for Family Health

Statistical models that consider the exhaustive categories of possible outcomes of child’s weight at birth are rarely found in literature. This paper, however, explores possible influence of household poverty level and maternal nutritional status on the weight outcome at birth of children under five years from data that were collected through multistage procedure. Weight at birth was measured on a five-level ordinal scale. Analysis based on traditional common regression techniques could not be employed for two reasons. Firstly, the response variable was observed on ordered categories. Secondly, household survey data often exhibit that responses of individuals that belong to the same cluster are correlated. Therefore, modelling techniques that take cognizance of ordinal responses and dependence among observations are suitable for correct inference. We analysed a dataset on child’s weight at birth from the 2003 NDHS using ‘svy’ command in STATA. This method produces more robust estimates of the standard errors.