There IS a certain element of belief in it. This comes from the fact that some theories support and explain the facts but are not directly, empirically testable, for any number of reasons. For instance, in the case of evolution, the long timelines prevent empirical testing, so we look for less compelling proofs of the theory. M-theory / string theory in physics is another example. Many of the predictions of M theory regard regions of tightly curved space whose dimension is on the order of the so-called Planck length, which is marks the lower limit of physical measurability due to consequences of Heisenberg's uncertainty principle. Theories of sociology or politometrics are testable in reasonable time but involved stochastic (statistical) elements that prevent the exact determination of consequences.
This does not make any of the above theories invalid, but it does allow sufficient uncertainty to leak into them that they necessarily involve an element of incompletely justified academic confidence, if not outright belief.
Thomas Kuhn's book, "The Structure of Scientific Inquiry" shows just how much belief goes into science. That said, science's redemption is its general ability to connect theories to consequences with the fewest and least elaborate steps in between. Systems of pure belief generally involve many MANY more assusmptions and outside influences on the phenomenological world than should be necessary to explain what we see, most of which beg further explanation.