Recent Center for Research on Education Outcomes (CREDO) analyses find that cyber charter schools in seventeen states show consistently low reading and mathematics value-added test scores compared to traditional public schools serving comparable students. This generally accords with prior research. We hypothesize that the relatively poor measured academic value-added of cyber charters reflects artificial testing conditions for students in those schools. Accordingly, we have collected testing information from the seventeen CREDO states. State-level analyses find that cyber student persistence, which likely indicates school quality, correlates moderately and significantly with the cyber student academic value-added as measured by CREDO. Further, we find evidence of lower cyber school value-added in states which permit cybers to use narrow testing windows, perhaps reflecting testing fatigue on the part of test-takers. We discuss implications, and suggest next steps for research exploring whether testing conditions affect measured cyber charter performance.