If a researcher wants to estimate a population proportion, and they have no idea what the proportion $p$ is, what value should they use for the sample size calculation?