Of course religious scientists aren't proof of compatibility, but you might think they're some evidence. In fact, you would almost have to think so, if, like Coyne, you regarded it as evidence of incompatibility that people lose religion as they develop into elite scientists--Some argue that the mere existence of religious scientists proves this compatibility, but that is specious. That people can simultaneously hold two conflicting worldviews in their head is evidence not for compatibility but for Walt Whitman’s (1855) solipsistic admission, “Do I contradict myself?/ Very well, then I contradict myself,/ (I am large, I contain multitudes.).” This argument for science/faith compatibility is like saying that Christianity and adultery are compatible because many Christians are adulterers.
Further evidence for incompatibility comes from the huge disparity in religiosity between scientists and laypeople. While only 6% of the American public describe themselves as atheists or agnostics, 64% of scientists at “elite” American universities fall into these classes (Ecklund 2010; similar results were found by Larson and Witham 1997). This figure is much higher for more accomplished scientists. A survey by Larson and Witham (1998) showed that that 93% of members of the National Academy of Sciences, America’s most elite body of scientists, are agnostics or atheists, with just 7% believing in a personal God. This is almost the exact reverse of figures for the American public as a whole.
This disparity bespeaks a profound disconnect between faith and science. Regardless of whether it reflects the attraction of nonbelievers to science, or the fact that science erodes religious belief—both are undoubtedly true—the incompatibility remains.
Why worry about "proof" in the first passage, and then lower the bar, making the issue "evidence" in the second passage? With evidence the topic in both passages, both the religious scientists and the formerly religious scientists come out to be some evidence. But it doesn't seem as if Coyne wants to accept the religious scientists as any evidence at all, since he dismisses them entirely with the Whitman quote and the Christian adultery point.
Now, if you've made up your mind that religion and science are logically incompatible (cannot both be all true) on some independent basis, then yes, you can dismiss Francis Collins as compartmentalizing (that's what you've got to think--or something along those lines), and you can think formerly religious scientists are reasoning when science gradually crowds out religion for them. But in that case, you've got some other basis for thinking religion and science are incompatible. What's really hard to do, if you're being fair and logical, is both dismiss religious scientists as any evidence at all for science-religion compatibility and also use formerly religious scientists as some evidence for science-religion incompatibility.
Now maybe, maybe, there could be special reasons for the different treatment. Maybe you've got evidence of compartmentalizing in the religious scientists. They may actually utter the words of Whitman's poem themselves! Maybe you've got evidence of reasoning in the formerly religious scientists--you've caught them going from premises about science to negative conclusions about religion. But barring special evidence of that sort, if you think scientists losing religion count as some evidence for incompatibility, you have to think scientists not losing religion count as some evidence for compatibility. Right? Right!