In our society, there are polls to show trends in just about everything. In fact, the most popular show on network tv these days is in effect, a popularity poll. Most of the polls that I see, however, are about politics. And where polls can be used and interpreted for a variey of things, many times, people, but more often than not, the media, totally misuses these devices.
Polls are not definitive. They all have "margins of error" meaning that if someone garners a 50% approval in a poll, the actual number could vary from 3-5% usually. this means that the person might only have a 45% approval or 55% approval in actuality.
Polls are a "snapshot." This means that a poll reflects a moment of time, usually a day to a week. Most polling takes place in just a few days to ensure that the "events on the ground ' are similiar for all the respondents.
Polls show trends more than they show actual numbers. this means that in order to accurately look at something that is being polled, it will usually serve one better to look at a series of polls that are identical in their questioning and polling source. For example, if you look at a poll of voter preferences from January of a given year till July, you can usually get a better feel of people's feelings and reactions to certain events beyond the "knee jerk" reaction of just 1 poll in that stretch. And where the actual number in every poll might be off slightly, the trend will better show where opinions and views are heading and what sticks and what doesn't in the news.
But the one poll that NEVER has any validity whatsoever is a national presidential poll. National presidential polls are about the most meaningless polls in existence. Especially in the political realm.
The biggest reason they hold absolutely no water is the fact that we don't elect president's by any national election. We hold individual state races that determine a state's electoral votes, which go to one candidate 100%. Meaning that if a candidate loses 1 state with 4 electoral votes by a 55 to 45% margin and another like sized state goes 90% to 10% the other way, the end result will be 4 electoral votes for each candidate. A tie. This is despite one candidate having a far greater share of the combined popular vote for the 2 states.
A national poll cannot ever hope to accurately reflect anything that has to actually do with the actual election. And i'm not just talking about the polling being done today, a year and a half out of the 2008 election.
Whereas some of the other polls will also not necessarily show accurate numbers necessarily, they can show trends. In a national presidential poll, the trends don't matter anymore than the individual polls. The only thing a national presidential poll can be really used for is morale. Either to crush the morale of a certain part of the electorate because they are "so far behind" in national polling or energize a particular group by showing a lead in national polls. A more sinister view could rationalize that the polls could be used in the converse fashion, effectively making one side or the other over confident or falsely demoralized if used as a propoganda tool by anyone.
In any case, where many polls are taken too seriously, they do have their place, if used properly.
Like many things, if you use something for a purpose other than for what it was intended, odds are that problems can occur.
But in the case of these national presidential polls, they have no purpose, can only confuse the public and are a complete waste of time to make, observe or study. "