This is not true with regards to the exit poll results reported by the US network news media for national elections. These are, in fact, funded by the networks themselves and run by survey researchers. For live election day polling, the voter precincts used are selected via random sample.

Of course there are several weaknesses in exit polling methodology as employed by the networks. But at its heart the methodology is statistically based, generally conforming to the tenets of standard social science quantitative research methodology. So, while not perfect, it cannot be said that exit poll results are shots in the dark either.