The distribution of consumer-resource body mass (CRBM) ratios has received increasing attention in the food web literature. In this study, we investigate how commonly seen right-skewed distributions of log-transformed CRBM ratios can arise. We identify two mechanisms that could account for these empirical patterns: (i) food web dynamics filter out interactions between consumers and resources wherein the CRBM ratios are less than 1 (consumers are smaller than resources), and (ii) Sampling biases prevent the detection of CRBM ratios <1. Using a generalized Lotka-Volterra model that uses allometric scaling to relate body mass to growth rate, we examine the expected distribution of CRBM ratios in assembled webs. We find that log-transformed CRBM ratio distributions that emerge from stochastic immigration and interaction-driven extinction dynamics are symmetric (approximately normal) rather than skewed. This holds across a wide range of underlying body size distributions of the regional species pool. We suggest that empirically observed skewed log-transformed CRBM ratio distributions do not necessarily reflect dynamical constraints, but are an artifact of sampling biases. The main source of such bias stems from practical difficulties in detecting interactions wherein consumers are much smaller than their resources (such as host-parasite and host-parasitoid interactions). We conclude that greater effort needs to be invested in quantifying food web interactions with CRBM ratios <1 (commonly observed in the form of parasite-host, parasitoid-host, or herbivore-plant interactions).