In a published randomized controlled trial, household units were randomized to a nutrient bar supplementation group or a control condition, but the non-independence of observations within the same household (i.e., the clustering effect) was not accounted for in the statistical analyses. Therefore, we reanalyzed the data appropriately by adjusting degrees of freedom using the between-within method, and accounting for household units using linear mixed effect models with random intercepts for family units and subjects nested within family units for each reported outcome. Results from this reanalysis showed that ignoring the clustering and nesting effects in the original analyses had resulted in anticonservative (i.e., too small) time x group interaction p-values. Still, majority of the conclusions remained unchanged.