How do you know that your discoveries are true (or likely true)? Do you use statistics or just common sense?
History research requires source analysis (primary, secondary, tertiary) and the interpretation of the source as evidence. Rarely is this directionless; typically the topic is rather narrow, and a very specific question about the topic is driving the research.
However, it is not enough to only consider the primary source. It is necessary to familiarize yourself with the historiography of the topic. This means reading the secondary sources (peer reviewed books and scholarly articles, referencing the original source), and researching the authors to check for credentials, bias, methodology, the contribution of contemporary intellectual movements, translation and other language skills, general knowledge of place and time, etc.
With proper source analysis, proper knowledge of the research and its theories (prevailing, fringe, or otherwise), and contextual knowledge of the place and time gleaned from study, a person can develop an argument based on evidence, but the process may lead to additional questions and additional research, and may require the development of language skills or consultation with many other historians with expertise in other time, places, topics, or even disciplines.
Historical research is a lot of reading, a lot of language work, occasional work with artifacts, studying historiography, and ultimately communication, peer feedback--repeat!
The narratives and arguments presented by historians tend to compete, with historians generally reaching a concurring opinion on the most likely scenario based on the evidence offered by sources. Some topics never reach consensus. Statistics can play a part, but they should be based on source evidence and come from a reputable source. Common sense plays a part, but an individual's judgement is subject to criticism, and rightly so! If you and I both look at the same primary source, we might not reach the same conclusion. What is considered to be fact is part of an ongoing, evidence based discussion in order to reach the most likely "truth."
For a more complete answer, consider reading the following:
Gunn, Simon, and Lucy Faire. Research Methods for History. Edinburgh: Edinburgh University Press, 2012.
Hoefferle, Caroline. The Essential Historiography Reader. Upper Saddle River, NJ: Prentice Hall, 2011.