Snapple wants to survey the general population to see what proportion prefer bottled ice tea over soda. They want to be 97.5% confident in their estimate and have an acceptable magnitude of error of 3.0 percentage points. Because they have never done such a study before, Snapple does not have an estimate of what the proportion might be. If a simple random sample is used without a population correction factor, what is the minimum number of sampling units needed?