The cost of a certain phone call was $0.75 for the first 3 minutes and $0.20 …

Correct answer is (B). 

This problem highlights how important it is to read carefully and to look for potential interpretation errors in GMAT math questions. It also shows how it is often easier to manipulate the question stems to match the statements in data sufficiency questions than to change the statements to match the questions (what you naturally want to do). Here is the incredibly well-made “con” in this question:
 
A majority of high performing students do not properly interpret how to determine the cost—the instructions say $0.75 for the first 3 minutes, not $0.75 per minute for the first three minutes. However, most people carelessly calculate the charge as if it were per minute. If you do that improper interpretation, then the question stem seems to be asking this in terms of cost: a 15-minute call would be 3 × ($0.75) + 12 ($0.20) or $2.25 + $2.40 = $4.65, so the question would be “Did the phone call cost more than $4.65?” Statement (1) would give you a definitive “No” to the question (cost would always be less than $4.65) and thus be sufficient. Statement (2) would allow for the cost to be both below and above $4.65, so the “Maybe” answer would make it insufficient. With the improper interpretation, you seem to have done everything correct when picking (A). But the correct answer is (B)!
 
The proper interpretation:
 
The first three minutes in total cost $0.75 and each minute after the first three costs $0.20 per minute. A 15-minute call would cost $0.75 + 12 ($0.20) or $3.15. So, after changing the question to ask about cost (in order to match the statements) it becomes: Did the call cost more than $3.15?
 
Now you see that statement (1) gives a maybe answer since it allows for both a yes and a no answer to the question. Statement (2), however, guarantees that the cost will always be greater than $3.15, so it is sufficient, yielding the correct answer of (B).
 
Questions that are designed to elicit one particular mistake are dangerous on this test: if you make that mistake, it will feel like you have gotten the question correct when you really have not. To avoid these “designer mistakes,” you must always be critical and try to find the trap or interpretation mistake in all GMAT quant questions. Additionally, this problem shows how it is often easier to change the question to match the statements in data sufficiency: it is much faster to change the time to cost in the question stem, than to change each statement to match the time in the question.