第一次写AA，软件限时模拟之下，请各位狠批！

gerrardliv8th

时间上貌似是够的，不过可能是因为没怎么想好就写了，感觉论述的有些偏题。。。大家看着办，千万别留情面。
（No. 82）


The author argues that since the voters keep changing their position before finally casting a vote, a forecast, which is based on opinion polls, is no better at predicting an election than a random guess do. However the statement omits substantial evidence to differ a forecast from a poll and fails to estimate the percentage of undecisive voters to the whole voters.


First of all, forecasts on presidential election will consider various factors such as candidates' policy, the current economic performance, the international environment in addition to citizens' opinion. Therefore, a forecast, though based  partly on opinion polls, will comperhensively present the possibilities of different candidates being the most favored one. And sometimes we can find that one of the candidates is forced out of the election beacuse of unjustified policy or sexual scandals. Random guess can not anticipate situation like that.


Secondly, even though voters change their opinions untill the last moment, they account for rather a small percentage in the public and thus have limited influence on the credibility and utility of an opinion poll. Empirical evidence has shown that most of the times an opinion poll can reflect people's preference on the candidates and even become the basis on which strategies of candidates' campaigns are modified. Simply, people won't take a random guess into account but an opinion poll.


Nevertheless, when two presidency candidates are having almost the same votes from the poll, the election outcome become more like a random guess when the relative small number of undecisive people can really make a change on the final result. However, the poll still predicts the touch-and-go situation while the random guess can have an unrelated result.


On the whole, forecasts based on opinion polls serve to be an effective tool on predicting the result of presidential elections while a random guess take nothing into account and consequently is not reliable.