Inferring from evidence is the process of drawing logical conclusions by combining available clues, data, or observations with background knowledge. It goes beyond explicitly stated information to fill in gaps, essentially "reading between the lines". Effective inferences are reasonable deductions, not random guesses or unsupported assumptions.

Richard Dawkins: I had said that evolution is a fact in the same sense that it's a fact that the Earth orbits the sun. And somebody wrote in about that saying "it isn't a theory the earth moves around the sun, it's a fact we can observe, evolution is not."

I see why he says that, because you sort of think 'well, the earth is moving around the sun all the time and therefore you can observe it, whereas you can't actually observe evolution cause it happened a long time ago', well mostly it did.

But really you can't observe that the earth moves around the sun. What you observe is the changing seasons, and you infer that the earth moves around the sun because you have an interpretation, you have a theory to account for why you get changing seasons.

In the same way you observe that night follows day and day follows night. You don't actually observe that the earth is spinning; you have a theory that the earth is spinning and it accounts for the facts. And it's just the same way with evolution.

It's true that we can't actually observe at least the major features of evolution happening. We can observe very very recent, very tiny bits of evolution happening. It's true that we can't watch dinosaurs evolving, and we can't watch fish emerging from the sea and becoming land animals. We have to make inferences. We make inferences on the basis of the evidence available. In just the same way as we make inferences about the earth's orbit by looking at the seasons and the earth spinning by looking at the day night cycle.

I have likened this to a detective coming on the scene of a crime after the crime's been committed. You can't actually watch the crime being committed, but you can see footprints, you can see fingerprints, you can see all sorts of clues that remain in the room where the murder was committed. And you can put forward a hypothesis as to how the murder happened, and then check the hypothesis against the facts. And to a greater or lesser extent, the detective can come to a firm conclusion about who the murderer was.

It's just the same way with evolution, except that the number of clues that we have after the crime so to speak are enormously more, enormously richer than those that any detective normally is privileged to see.