<p>Model selection approaches are gaining popularity in biological research due to their utility in evaluating support for multiple candidate hypotheses. However, top-ranked models from a set of candidates do not necessarily describe the underlying processes that give rise to biological phenomena or provide strong predictive ability. The field of invasion ecology is increasingly using comparative functional response (FR) approaches to predict the trophic impacts of invasive species based on the FR model that best fits experimental data. However, noisy experimental data and a variety of, at times, conflicting model selection approaches may limit the ecological interpretation of results. Here, we use experimental (empirical and simulation) and analytical approaches to explore how the ecological interpretation of FR data can be obfuscated by methodologies (i.e., experimental habitat complexity). Finally, we survey the literature and identify which model selection approaches are most common in FR experiments in invasion ecology, and how the resulting model fits are interpreted. The round goby is a prolific invasive fish in North America, responsible for local declines in invertebrate populations through predation. Using round goby as a model predator, we demonstrate that prey-type (mobile versus immobile) can shift the best-fit FR from Type III to Type II. In seven out of eight empirical treatments of varying habitat complexity, and eight out of eight corresponding simulated treatments, model selection outcomes differed depending on the analytical approach used. Our results demonstrate the context-dependence of FRs and highlight the limitations of these FR experiments and associated model selection methods. We encourage researchers to critically assess model selection methods and results when identifying and using top-ranked models, and provide recommendations to improve predictive accuracy.</p>