The overuse of test accommodations (e.g., test readers, extra time, and calculators) for students with disabilities is a potential means of gaming the accountability system because it can inflate proficiency gains. However, no direct evidence on this problem exists, and findings on whether or not test accommodations improve test scores are persistently mixed. A key issue underlying both problems is the failure to account for contextual attributes. We propose students' skill range--the breadth of knowledge and skills they acquire--as a fundamental contextual variable. A wider skill range implies a broader and thus more complex skillset. Under such conditions, more students with disabilities are likely to legitimately need test accommodations. Gaming is thus suspect when a high percentage of students with disabilities are given test accommodations and the skill range is narrow. Based on state-level panel data (2007-2009) for eighth graders, we find that states with a wider mathematics skill range indeed use test accommodations more commonly, and that under such conditions test accommodations do not necessarily result in greater proficiency gains, suggesting that gaming may be less likely in such states. Proficiency gains from test accommodations are greatest in states where the mathematics skill range is markedly narrower, yet a high percentage of students with disabilities are nonetheless given test accommodations. We offer a number of competing heuristic explanations of gaming and no-gaming under a wider set of combinations for skill range and test accommodations, which are testable in future research. We follow this with a discussion of broader implications for scholarship and practice.