The concerns you note are handled by the processes of peer review and attempted replication. These are strengths of the scientific method not weaknesses. Additionally, statistical confidence intervals are clearly stated, often in the abstract.
It's not a perfect system, no matter how you juggle the math.