I try and make a habit of reading actual scientific papers. Science journalism, while providing a valuable service, often exaggerates or loses the point, or fails to contextualize properly. With Google Scholar, SSRN, arXiv, PubMed and many more, this gets easier and easier (no thanks to companies like Elsevier).

Although there are some problems with reproducibility - or are there? -  I still like do keep this habit up (even though Links to the Damn Paper isn't updating anymore).

One of my favorite sources is the JASNH: Journal of Articles in Support of the Null Hypothesis

[PDFs ahoy!] What's not to like about finding that the phases of the moon aren't linked to births or deaths, that questionnaire color doesn't affect feelings about the environment, that jackdaws don't remember things in order, that culture doesn't affect Olympic performance, or that women do not systematically prefer a man in uniform.

One of the biases we confront every day is one of events. We notice and remember things that have happened, not things that haven't happened. It's difficult to measure non-events, but the JASNH helps. And sometimes - a lack of information is still information.

CCBY-NC-SA4.0