It's rarer still for questions to be released that turn out to have problems. In most cases these issues are extremely subtle, and sometimes turn on disputed points of grammar. The flap about a question on a 2003 PSAT, for example, involved a pronoun use that a few handbooks condemn, even though it's standard usage among most good writers.
This morning, however, while analyzing questions on another old PSAT, I came across a question that I believe is truly wrong. Such an event is sufficiently rare that I spent a long time second-guessing myself to see if I was somehow missing something, but I don't believe I have.
Here's the question, which comes from the Wednesday form of the 2004 PSAT:
The newscaster (A) reported that (B) this winter consumers can expect the price of fresh produce (C) will increase (D) more rapidly than the price of meat. (E) No error
The keyed answer is C. I believe there is actually no error and the correct answer should be E. Fortunately, we don't have to speculate as to why the College Board believes that C is an error. This same question appears in something called the "12th Grade Premier Agenda," the answers for which are available on the College Board web site. The explanation given there is that "In this context, the verb 'expect' requires an infinitive ('to increase') as its complement, not a finite form of the verb ('will increase')."
As College Board explanations go, this one is fairly clear, up to a point. Unfortunately, it's supported neither by real usage nor by any usage guide I can find. This explanation amounts to an assertion that it incorrect for expect to be followed by a content clause, and as this sentence is written, that's exactly what "the price of fresh produce will increase more rapidly..." is.
If the College Board explanation is right, then the second of these is wrong:
Infinitive: I expect him to return the book.
Content clause: I expect (that) he will return the book.
I don't buy that claim. All the evidence indicates that, while the form with the infinitive is somewhat more common, both are entirely standard. Google's Ngram viewer shows that expect to is more common than expect that, by about 2-to-1, but instances of expect that amount to a little under 10% of the total over the period 1800-2000, and relative usage does not really shift in any significant way. In other words, expect + a content clause can't be dismissed as an innovation. Moreover, this search only shows content clauses introduced with that, and so the total percentage of content clauses must be higher.
With more targeted searches, it's easy to find plenty of examples of writers with impeccable literary credentials using expect + a content clause, both with and without that. For example, here are two from Samuel Richardson's Clarissa:
I expect that you will break the seals of this parcel, and when you have perused them all, give me your free opinion of my conduct.
And now, my dear, what shall I conclude upon? You see how determined—But how can I expect your advice will come time enough to stand me in any stead?
And here's one from Jane Austen's Emma:
This is in the supposition of his attachment continuing what it now is; but I do not know that I expect it will; I do not look upon him to be quite the sort of man - I do not altogether build upon his steadiness or constancy.
Usage guides don't complain about expect + a content clause. The Merriam Webster Dictionary of English Usage has an entry for expect, but it is largely devoted to the (now obsolete) controversy about whether it's acceptable to use it in the sense of "suppose," a point that is irrelevant to this sentence anyway. It does mention that it "is very often followed by to and the infinitive," however that assertion is a long way from claiming that a content clause is wrong.
Finally, ordinary dictionaries routinely give examples of expect (that) with no indication that they are non-standard. See, for example, here.
In short, expect allows two equivalent patterns of complementation with essentially the same meaning. Even though the infinitive form is more commonly found, it's certainly not required in this context
Since this so-called error isn't prominent in usage handbooks, I can only imagine that the question writer followed his or her intuitive sense of what sounded off, and in the writer's internal grammar, "to increase" sounded much better. The only problem with questions about idiomatic usage is that your internal sense, no matter how educated, cannot substitute for the overwhelming evidence of the use of other people. Such questions are hard to write for just this reason: instinct can be idiosyncratic, and a fair test should not impose the question writer's idiosyncrasies on test takers.