Just like scientists, we develop hypotheses to try to explain our historical evidence (but we usually call these 'arguments' instead of hypotheses), and then we test these hypotheses against historical data ('sources'). If the data contradicts our hypotheses, we try to find arguments that work better; if the data seems to support our idea, we publish it for peer review, critique, and further research.

Where we differ from the hard sciences is that it's very difficult to isolate variables in historical study. Because the past is gone, we can't experiment on it. We have to settle for whatever evidence survives into the present and use that to test our arguments. This evidence is almost always extremely messy, and cannot be reduced to single controlled variables. Without controls and the ability to isolate variables, the tests we use to explain hypotheses are always going to be very unreliable. Add to that the fact that human society (unlike, say, a chemical reaction) is immensely complex, and you have a lot of uncertainty to grapple with even when you can pick and choose your data (like sociologists working in the present are able to do).

Historians, as a profession, spend most of our time cleaning up our evidence into a form that we can begin to use to create and test arguments about the past. History PhDs take nearly a decade to earn because so many tools are required to make sense of this messy data.

For example: sociologists use statistics to study data from living populations; many historians do this too, say with 19th century census data. But before we can use statistics on this data, we have to first learn how to read handwriting written in another language; then we have to learn how census data was gathered 100 years ago in that culture, and how that might bias the accuracy of the data (which requires collaboration with other historians who have studied census gathering practices in the past); then we analyze the data; and then we have to contextualize this data within everything else we know about this period. And all that other knowledge has its own host of interpretive issues we have to understand before we can understand it -- each of which is an expertise of its own.

Historians have to be familiar with some combination of literary analysis, statistics, archaeology, art history -- the list goes on. We use each of these skills to try to understand what a specific piece of the past means within its own time and place, and then we assemble all this together into a set of data that, if we've done our work well, will allow us to start to test our arguments. This is all a way of isolating as much of our 'variable' as we can. It's a messy process. But then, so is most of what real scientists do, it's just on a different scale.

Our conclusions aren't infallible, but they can be checked by other historians who reassess the same data or, more often, create new sets of data from analyzing other kinds of sources which show a different (sometimes contradictory, sometimes complimentary) side of the story. We publish, we peer review, and we revise our arguments, adapt, and work toward consensus. In this, too, we resemble the sciences: we're always updating our ideas in response to the work of our peers as we strive toward arguments that explain as much of the available evidence as possible.

Are there any events where the truth is still in contention? Oh yes! There are some things that are very well established, because they've been studied exhaustively. We know that the Holocaust really happened, for example, and we agree on the major details. Other things are harder to answer: we still argue, for example, whether women fought in armies in the early middle ages, and we will continue to do so for a long time as the evidence is very fragmentary, which leaves a lot of room for subjective interpretations (and when our evidence is bad or inadequate, we're often stuck with it, unlike the hard sciences--we can't do a new experiment to generate cleaner results). Some things are difficult to pin down because there are too many variables to isolate: the cause of WWI, or the fall of the Roman Empire, can never be explained by a single theory because there are too many moving parts.

In the end, just like the sciences, it's easier to say what didn't happen than what did. This is valuable -- these counterfactuals help us narrow in on the truth, and they also help us understand what could have been but wasn't. Since historians' ultimate goal is to understand what it means to be a human living in a world that changes through time, knowing what might have been but wasn't, and what probably actually was, helps us find our place in the world.

---

In practical terms, we read a lot of books.

But the methods we use to turn these books into evidence that can be analyzed, tested -- that requires our specialized training, and these methods are what separate everyday opinions from the tested, expert conclusions of rigorous historical analysis.