This is something I’ve written about before– people dramatically overestimate the sample size needed to make responsible statistical conclusions.
It depends on what and how you are studying it how large your sample size needs to be – simplifying things greatly, when reading papers confidence interval, confidence level and standard of deviation are what matters.
To simplify things even more, you don’t need a very large sample size to determine with 99% confidence that decapitating a human will result in death because the effect size has little randomness.
What would that need, a sample size of five?
Anyway, more seriously, most people woefully overestimate the actual sample size needed for 95% confidence.
For instance, for determining with 95% confidence something with a variance of 0.5 and a confidence interval of +-5, you’d only need ~384 subjects. (Of course variance in many cases can only be determined after the study is done.)
So when someone claims, Oh, that can’t be right, they only interviewed 500 people! Well, no. 
For most purposes, 500 people is a whole fucking lot. It’s a huge sample size in almost all common cases.
Where studies and people often run into problems is that if the effect size is tiny, then even very, very large sample sizes do not produce valid results. (And beyond a certain point increasing sample size does very little.)
But that is a post for another time.
Note that I am not an expert in any of this, but I do read a lot of scientific papers and like to understand what I am reading well enough to have some sort of thoughts about it.