I've typed as ESTJ on more than a few occasions (including a three-choice quiz I took just now), especially the first few times I took it, which could suggest that I am an ESTJ who has mastered the implications behind the questions in order to manipulate my results towards ENTJ (however, this tactic is in itself unlikely to be very sensorish).
As I said in another thread, my instinctive feelings are that a lot of S types are no doubt confused about their type, and that they are actually N types. A lot of these people are "proud" that they are in fact an S rather than an N. (However, I'm sure a) there are some Sensors on here and b) some such Sensors recognise that their pride is well founded as they are surrounded by iNtuiters and are ironically more unique than iNtuiters on forums.) These so-called Sensors are essentially "victims" - for lack of a better term - of poor question phrasing or testing.
Take the ENTJ. Very few ENTJs would consider themselves head-in-the-clouds people more than they would down-to-earth people, and so what happens is they tick the Sensor box, scoring a point against their being iNtuitive. Alternatively, concrete vs. abstract: it depends entirely on the meaning of the term abstract. I hate a lot of philosophy; I find it very pointless and impractical. So would many ENTJs. And yet the test determines the iNtuitive type to tick the abstract box. If an ENTJ were to do this enough, they would come out as ESTJ. I've done this myself a few times. I'm down-to-earth, practical, and concrete. But this does not make me an ESTJ.
Generally then, I think the S/N distinguishing questions are the most flawed. The E/I, T/F and J/P ones are relatively straightforward and most people don't have an issue with answering them. But when it comes to S/N, I think those who struggle to answer them are probably iNtuitives, because questioning their very meaning in the first place suggests that they are in fact iNtuitive thinkers, looking for implications, principles and hidden meanings as opposed to literal meaning.