MODELING COUNT DATA WITH OVER-DISPERSION USING GENERALIZED POISSON REGRESSION: A CASE STUDY OF LOW BIRTH WEIGHT IN INDONESIA

Poisson regression is commonly used in modeling count data various research fields. An essential assumption must be met when using regression, which that the of response has mean and variance equal, namely equi-dispersion. This often unmet because many for greater than mean, called over-dispersion. If model contains over-dispersion, then will produced an invalid can under-estimate standard errors misleading inference parameters. Therefore, approach needed to overcome over-dispersion problem regression. The generalized handle study aims obtain factors affecting low birth weight Indonesia 2021. result shows based on were: poverty rate, percentage households with access appropriate sanitation, pregnant women at risk chronic energy deficiency receiving additional food, who received blood-boosting tablets, antenatal care.