Probably not that random. I imagine uf you answer one of these polls, your phone # goes on a list and you get called back more often.
I just spent some time going through the Fox poll internals, looking at the devil in the details. I found they not only over sample Democrat registered voters to weight the overall results against Trump, but when you look at the details to see how they got such high numbers of people wanting impeachment, you find they also skewed the sample to higher numbers of women, blacks, black women, and other minority women, especially inner and urban and suburban women. For example they were tracking “suburban women” who swung heavily (D), but hey no had category for “suburban men”. Odd. Minority women were largely anti Trump. . . and overly weighted in the sample. These were the thumbs on the scales. You have your callers keep calling, discarding calls until you get the demographic mix your seeking. . . The one you already know how they’re going to lean. “No, drop that call data, we already have enough rural rednecks who like Trump. . . That quota is filled. Keep calling until you find a Hispanic suburban housewife or an urban male transsexual. We don’t have enough of those yet. . . oh, and we need some atheist gay Republicans. C’mon, chop chop. Keep calling!”