How can scientist tell doing/taking one thing or not will make you live longer/less?

Probability, haven't you seen that exhibit at the science musseum were 100 balls are dropped randomly in a glass box, they fall in a predictable pattern.

If the average person dies at age 82 and the average smoker dies at 47 it can be safely assumed smoking will have a negative affect on your health.

Extensive double-blind placebo controlled studies, stage 4 of the FDA drug approval process (long term studies), meta analysis of data, the process of peer review, etc. For many conditions, the end goal is clearly stated in the research. Ex:

-an increase in overall (OS)

-an increase in disease free survival (DFS)

-an increase in (x) year survival rate

*This is all in comparison to the survival rates of an untreated (or alternatively treated) placebo group. It is an imperfect system, but one that is far more reliable than the anecdotal evidence which healers have relied on for centuries.

If something permanently impacts your health such as smoking, it is safe to assume, although not guaranteed, that you will live less than someone who has the same lifestyle, but does not smoke.

Well, if you do a lot of cardio everyday, not only will you live longer, but the quality of your life will vastly improve

they can't , most of these claims are bogus and based on zero evidence or VERY shaky statistics