Shortly, because if someone believes in a single incorrect thing (also a quite big one that can't be overlooked), it increases the probability of him believing in other incorrect things.
The long explanation is that if the author believes in religion (something that's wrong - I know I might be wrong about it, but for reasons too long and irrelevant to write, I act as if I was certain that God doesn't exist), not only that he had tens of years of exposure to information and that didn't change his mind about it, but he actually believed in religion in the first place - following a conclusion from premises that don't lead to that conclusion. This is a (not necessarily strong) counter-indicator of ability to perceive the truth (I'm not sure if "critical thinking" is a synonym to that). Although it is entirely possible that he is in fact absolutely right, the expected outcome (I can't explain "expected outcome", I even started a thread on that - I trust on the reader to know what is it, otherwise the debate usually cannot happen) is that he is more likely to have flaws. Especially in a subject where he might follow anecdotal evidence and neglect the importance of scientific approach (some subjects might be more intuitive).
Of course, if I've read the book, all the information about the author becomes irrelevant, because it only serves to predict the content of the book. However, reading the book requires many hours (especially in my case - I currently desperately need time, and I tend to be inefficient in it, even when I'm not procrastinating).
P.S. Sorry if you can't understand how I use expected value to get to that conclusion, I just cannot explain it intuitively and I even (small) doubts about the validity of that approach, because I'm yet to see evidence of it.
I think you're likely overweighting this, at least in the general case.
It's hard to overestimate how good people are at selectively interpreting, and more importantly compartmentalizing, evidence to fit their identities. Now, selective interpretation alone would support your line of thinking -- if people accept only those data points that fit some preconceived notions, then of course their opinions aren't good evidence for anything related to those ideas, and religion theoretically touches just about everything. But when you take compartmentalization int...
I'm considering reading the book by the title How to read a book. A friend of mine (his critical thinking is quite good, but certainly not as good as it could be, so I can't trust his opinion too much) said he has read it and that it helped him a lot. He said it had advice on reading comprehension, critical thinking ("don't automatically accept what you read") and that when people read something, they tend to forget it quite easily (and that the book addresses this issue). But he also quoted a part of the book, which said that only reading hard things will improve your reading - it might be true, but it doesn't sound intuitive to me (according to my rationalist intuition, obviously :D). Also, the book is written in 1940 and revised in 1972. Additionally, the author is religious (I think he's even highly religious). And if I remember correctly, it's not based on research - there is a quite high chance that I don't remember correctly. I checked its Amazon page, nothing said anything about research (browsed through all the low ratings to see if they complain about that, nobody did).
Should I bother reading it? If it delivers what it promises, it will obviously be so cost-effective that most rationalists should abandon reading whatever they're reading and switch to this book. But is there a version that is entirely based on research, with references or sound theory behind most claims?