Diverse samples are valuable to the study of development, and to psychology more broadly. But convenience samples--typically recruited from local populations close to universities--are still the most widely used in developmental science, despite the fact that their use leads to a vast over-representation of Western, White, and high socio-economic status participants in our studies. Do convenience samples still have a place in our research? While diverse samples are crucial to advancing developmental science, policies designed to encourage recruitment of such samples may not always succeed in improving sample diversity in ways that will benefit our theories and reduce bias. Further, convenience samples are just, well, convenient--and because their costs are lower, they allow for faster and more precise research. We suggest three paths forward to resolve this tension: 1) use theory and observed variation to choose aspects of diversity to prioritize in a particular study; 2) use online methods as an important route to broaden which samples are convenient; and 3) work in teams to achieve "inconvenient" samples using many different convenience populations.