Hi, anyone could help with this: we know that with a considerably large sample we can estimate what is the proportion of cases having a certain characteristic and calculate the prediction interval of this percentage. The formula used for the interval calculation is conf.int = z * standard error of p.
Does anyone know what happens in cases where a certain characteristic might have very big values (eg 98-99% of the sample) and as a result the upper conf.int. goes beyond the 100% ? Can we just say that we round the number to 100% or is there any special approach? Thanks a lot
Does anyone know what happens in cases where a certain characteristic might have very big values (eg 98-99% of the sample) and as a result the upper conf.int. goes beyond the 100% ? Can we just say that we round the number to 100% or is there any special approach? Thanks a lot