Prior to an election, a market research firm is hired by a political party to do some polling. The firm is asked to gather some data so that the party can draw some inferences about the proportion of voters who would vote for them in the next election. On the basis of the firm's research, the party would like to be able to make a statement such as: "According to a recent opinion poll, 62% of all voters would vote for our party. This poll is in error by +/- 3% in 95 samples out of 100."

How many voters should the polling firm include in its sample, if it has no idea about the current estimate for the party's popularity with the voters?