PT139 · S4 · Q7 Most respondents to a magazine survey

Although we could strengthen this argument by saying, "People experienced both the taped version and the written version and almost always preferred the taped version", it doesn't weaken the argument to learn that most people did not experience both versions. This answer can't be said to much impact, since we basically already knew this to be true. The background facts of the survey made it clear that most people did not experience both versions, since most who listened said they enjoyed the novel and most who read said they didn't enjoy it. If most respondents had listened to it and read it, then we would be able to say "most of the listeners said the novel was good" and "most of the readers said the novel was bad" because most of the listeners were also readers (and vice versa).

This presents a distinction - reading takes less time than listening. Could that be the real reason why the listeners liked the novel more than the readers? For us to think of that as an Explanation, we'd have to assume that "the less time it takes to experience a novel, the less likely you are to enjoy it". That's not really a common sense link (for myself, the opposite probably applies). Furthermore, that wouldn't be an Alternate explanation; that would still be reinforcing the author's explanation. Since we can read faster than someone can speak, it is an inherent property of written books vs. taped books that the former will take less time to get through. So if we're attributing the positive/negative evaluations respondents had to how long it took them to get through the novel, then we are agreeing with the author's conclusion that the format itself will tend to have an influence on how much someone enjoys the novel.

This answer applies uniformly to both written / taped versions of novels, so it doesn't help us judge any distinction between written and taped versions. The fact that in both cases people are more likely to give a positive review than a negative review doesn't affect anything in this conversation. We might have thought, "Wait, in this survey, when people were asked their opinion of a best-selling novel that they have read, they were more likely to say they did not enjoy it than that they enjoyed it". Doesn't that conflict? Sure, it doesn't conform to the generalization, but the generalization wasn't guaranteeing that in all cases when you ask people about a best-seller they read or heard, there will be more favorable than unfavorable votes. It's just saying the probability is higher for positive reviews than for negative ones. We can't weaken by saying that one data point doesn't conform to a probabilistic statement. If we say, "Lebron usually drives to the basket with this left hand", we're not weakening that by pointing to one data point when he drove to the hoop with his right hand.

It doesn't matter whether audio versions always exist. The author is promising that we can find an audio version of every book. She is only concluding that if there is an audio version, we're more likely to prefer it over the written version.