Politics, An Irrational Science
One thing that continually befuddles political analysts, wherever they are, is the confounding behaviour of voters. The United States has politics down to a science (so their political scientists might claim), but examining their performance in the past two presidential elections, one would think this was anything but a science.
One thing politicians learn very quickly if they do not intuitively know it already is that people are swayed by their emotions, not their thoughts.
Your goal in a speech is thus not to appeal to their rationality, but to their irrationality — to make them feel, not think, that you are the best candidate.
Economics frequently assumes that all individuals are rational actors, calculating the costs and benefits of each action to them and behaving accordingly.
We know that this is not true, and many economists — most recently Alan Greenspan — have suggested that a more accurate view of individual actions ought to be taken.
The interesting thing about economics, however, is that its assumptions often work because they usually hold true. For a substantial portion of the time, some assumptions fail, but in the big picture, these deviations are not statistically significant to the point that they make most economic analyses meaningless.
The same cannot be said for political science. Economics at least gets people to quantify their confidence in something through the price system. They can count on some measure of honesty because money makes you literally, well, put your money where your mouth is.
Political scientists cannot conduct such a confident assessment of individual preferences; it's not exactly legal to poll people and get them to bet on a particular outcome.
Ironically but unsurprisingly, perhaps the most accurate predictor of election results has been online gambling websites. Although obviously they do not get every result right, far more often than not, they are right on the money.
It is a bit of a misnomer to label political science a science. A science is capable of developing a hypothesis, testing that hypothesis, and using it as a basis for future predictions. This is something political science has never been able to do; you can take the same people, with the same opinions, today and six months from now, and you can have a completely different election result each time.
Political scientists ought to acknowledge the irrationality of voters, and be honest about what political science is — it's not a science, it's an art of analysing and guessing, nothing more.