Evidence for a model (or belief) must be considered against alternative models. Let me describe a neutral (and very simple) example: Assume I say I have Extra Sensorial Perception (ESP) and tell you that the next dice throw will be 1. You throw the dice and I was right. That is evidence for my claim of ESP. However there's an alternative model ('just a lucky guess') that also explains it and it's much more likely to be the right model (because ESP needs much more assumptions, many of those in conflict with accepted facts and theories). This is a subject of statistical inference. It's crucial to consider the alternatives when we want to put our beliefs to the test.