I am not convinced with B being the answer.
It’s important to remember with Inference based questions that we are looking for something that we 100% know is true. Hence, checking to see if there’s even a possibility to falsify any of the others helps eliminate each of those options.
(A) – Consumption and profits are not related. It is very plausible that consumption may go up, but cost price could go up as well, leading to lower profits. Hence, we cannot claim 100% that profits went up. For the same reason, (D) can be eliminated as well.
(C) – Although total consumption of cigarettes went up 3.4%, we cannot make any verifiable claims on who smoked these cigarettes. Lets assume for a second that 1000 cigarettes were smoked by 50 people in 1973. If that went up to 1034 (3.4% increase), it is both possible that the number of smokers went up to 60, or went down to 3 (There is no data on how many cigarettes each person smoked). Hence, we don’t know anything about the proportion.
(E) – We know nothing about people’s tobacco consumption habits. Sales of chewing tobacco increased, but as stated with (C), we don’t even know if the numbers using chewing tobacco went up or down, much less if there were people who switched from cigarettes or not.
(B) on the other hand only makes a claim about per capita cigarette consumption (which is calculated by CigsConsumed/TotalPop). Since we know that population increased more than the cigarette consumption, per capita consumption must have gone down.
What works in favour of (B) is that it doesn’t state anything outside what the data already concerns. Very often, the challenge with an inference question is that it demonstrates to you that you can actually make very few verifiable claims with almost any dataset, and in day-to-day reasoning, many of the “natural assumptions” we make don’t fly on the GMAT