It’s not so easy to get the data! And there will always be aspects that are hard to measure, either because they are very rare or because it’s hard to design “observables” that tell us about the theory. I’d put it the other way around though: we make predictions because we’re trying to understand the world, and *then* we go out and search for data to confirm or deny the theories.
Making predictions is important because getting data is expensive 🙂 If we could do our work without experiments then no-one would do experiments. If you can get a model that predicts what happens in real life, then this is really powerful. But with all theories and simulations, experimental data is required to compare and validate these models.
I can’t easily get the data! Or at least, I can get computer data but not real experimental data.
I’m looking at reactor designs to see how they might behave in real life. It’s tricky to get experimental information because building a nuclear reactor is expensive, can be dangerous and takes a long time.
Instead we use a computer simulation, which is less accurate as we make a lot of assumptions, but it helps us get rid of designs that look really wrong!
Like other people have said it is not always easy to get data. The radioactive isotopes that I am studying are difficult to produce so we don’t get to perform experiments on them very often. I use simulations so we can predict what we will see and try to optimise the experiment so that when we do it for real we get better results.
Predictions are a crucial part of how science is done. First there’s a theory or hypothesis. This theory is used to make predictions. Then we carry out an experiment. If experimental data all disagree with a theory’s predictions, that theory can be ruled out as not representing reality. If experimental data agree with a theory’s predictions, we can keep using that theory until an experiment disagrees! Experimental data is the most important thing after all, since that’s the real world.