It is true that each of the answer choices is conceivably correct. But it is important to keep in mind that you are looking for the best answer, which will be the one word, which, more than all the others, is likely to be abused. As for (1), while there may be different ways of doing a random selection, we should be able to decide whether sample was in fact, selected fairly. Although the ad may be lying about the selection of participants in the study, we should be able to determine whether they are lying. In other words, though they may not have selected the sample randomly, they cannot escape by saying, “Oh by random we meant anyone who liked the Santo.” The same is true of (3), “first” that is a fairly clear term. You add up the answers you got, and one will be at the top of the list. Now, (2) is open to manipulation. By asking our question correctly, that is, by finagling a bit with what we mean by handling, we can influence the answers we get. For example, compare: “Did you find that Santro had a nice steering wheel?” “Did you find the wheel easy to turn?” We could keep it up until we found a question that worked out to give a set of “responses” from “randomly” selected drivers who could rank the Santro “first”. Now, if one category itself is susceptible to manipulation, imagine how much easier it will be to manipulate a “Composite” category. We have only to take those individual categories in which the Santro scored well, construct from them a “composite” category, and announce the Santro “first” in the overall category. There is also the question of how the composite was constructed.