Interpreting margin of error estimate

Question

Take this as a thought experiment at first, although it relates to something we are doing through our nonprofit: Assume every member of a population (not a sample!) of 550,000 is asked a yes or no question. 150,000 of them respond 'yes' or 'no,' the rest don't answer. How close is this response to that of the entire population, had they all responded? At a 99% confidence level, these numbers give a margin of error of about 0.28% (I think!). Can we interpret the margin of error to mean that the proportion of yeses and noes in the data set of 150,000 voluntary responses will be within 0.28% plus or minus of the 'actual population response,' 99% percent of the time?