There are two reasons why it is very difficult to test macroeconomic theories.
First, it is impossible to run experiments. Let us say that you had a theory that the Bush tax cuts prevented the economy from growing as quickly as it should have in the early 2000s. You cannot go back and run the economy again, only without the Bush tax cuts. Macroeconomic theories play out in the real world and cannot be replicated experimentally.
Second, there are too many possible confounding variables in the real world. This blurs the causal connection between a given macroeconomic policy and the conditions that the policy might cause. For example, let us say that you think that the economy would have recovered more quickly after 2008 if President Obama had not pushed through his economic stimulus plan. You could point to the slow growth we have had since then and say that that proves that his plan was a bad one. But what if there were other factors that actually caused (or helped to cause) the slow growth. What if it was really the Euro zone crisis that caused slow growth? What if it was the fact that the Republicans and Democrats could not agree on policy and so people were unsure about the stability and wisdom of our government?
There are so many variables that can have an impact on something as big as the whole economy (which is what macroeconomics studies) and we cannot run experiments to isolate the impact of any one variable. Therefore, it is very hard to test macroeconomic theories.