How much data is enough to reliably measure individual differences in cognition?