This is a binomial distribution since the answer to the question is yes or no, which is a two-state answer. However, a reply of no doesn't necessarily mean the voter has voted for party B, since A and B are the major parties, implying that there may be other minor parties. So if the purpose of the survey was to find the distribution of voters for each party, it would not be conclusive. Better statistical results would be obtained if the random voters were specifically asked which party they voted for. If the purpose of the survey was merely to estimate how many out of the electorate voted for party A, regardless of how many choices of party there were, then the binomial distribution would apply.
The results of the survey would establish the balance of voters in the sample of 600 from which can be deduced p, the probability of a random voter voting for party A and (1-p) for another party.
Because 600 voters is a sample, in a population of 7500, the results of the sample survey (8% of the registered voters) need to be measured according to the confidence level when the results are applied to the electorate. The quantity np where n=7500 gives the expected mean, np(1-p) the expected variance and the square root of the variance is the standard deviation.