This is a fair question. In science we take random samples, right? So do pollsters and focus groups, etc. If the 100 employees were chosen randomly or at least semi-randomly (self-selected groups like this can never be truly random), and almost all of them reported negative feelings toward the work culture, then yes that would be statistically significant. But if the New York Times was asking for people who hated their jobs, you’re totally right that we might just be seeing a tiny sliver of the truth. It all depends on how the NYT found these people to interview.