Often it's difficult to get good measurements on phenomena (eg on a specific battle in a war before the advent of freely searchable, practically real-time satellite imagery) which may mean you have to seek out reports from first-hand witnesses/participants/victims. In those cases, I advise searching for source documents from first-hand sources/legitimate researchers and available source footage, and if you can authenticate the material, you can use it to guide formation of an understanding. There's a lot of uncertainty in this and it takes time, but if you really want to know an answer you have to put the time in.
Obviously it isn't viable for you to dig into every single thing, so I'd say my shorthand method is to lean towards experts who have access to source data/data generating processes of interest and read their papers.
Regarding trusting news sources: all news sources misinterpret some things, but not to the same degree. The editors and journalists are the real QA mechanisms, so use the past accuracy of the company and the journalist in determining the baseline amount of uncertainty you should have in any specific piece of reporting.