Biological hypotheses predicting patterns of offspring begging typically concern the covariance with hunger and/or development at specific hierarchical levels. For example, hunger drives within-individual patterns of begging, but begging also drives food intake among individuals within broods, and begging and food intake can covary positively or negatively among genotypes or broods. Testing biological phenomena that occur at multiple levels therefore requires the partitioning of covariance between traits of interest to ensure that each level-specific relationship is appropriately assessed. We performed a partial cross-fostering study on a wild population of great tits (Parus major), then used multivariate mixed-models to partition variation and covariation in nestling begging effort and two metrics of nestling hunger within versus among individual nestlings and broods. At the within-individual level, we found that nestlings begged more intensely when hungrier (positive correlation between begging and hunger). However, among individuals, nestlings that were fed more frequently also begged more intensely on average (negative correlation between begging and hunger). Variation in nestling mass did not give rise to the negative correlation between begging and hunger among nestlings, but we did find that lighter nestlings begged more intensely than their heavier biological siblings, suggesting that this effect may be driven by a genetic component linked to offspring size. Our study illustrates how patterns of covariance can differ across biological levels of analysis and addresses biological mechanisms that could produce these previously obscured patterns.