The key in this question is that we have to see what skews the proportion of people that are dishonest. We have to note 'the proportion of job applicants who are dishonest' and NOT how dishonest one is. Whether one is a little dishonest, medium-dishonest or very dishonest, HE IS DISHONEST.

By saying that two-fifths admitted to being atleast a little dishonest, the statement implies that there could be more people that are dishonest.

(A) If some dishonest people have claimed to be dishonest, the actual fraction of dishonest people are more than two-fifths. This must be resulting in a wrong estimate of the proportion.

(C) talks about the degree of dishonesty. Even if those who have claimed on the survey are at least a little dishonest may be very dishonest, it will not have any effect on the proportion of people who are dishonest.

Say there are five people in the survey: A, B, C, D and E.
And, say A, B have claimed to be atleast a little dishonest.
If between A and B say B is actually very dishonest, the number of dishonest people will still be 2 (A and B). So, there is no effect on the proportion.

The survey results say that two-fifths of applicants are dishonest. There is no mention of levels of dishonesty: slightly dishonest, very dishonest, etc. It's an either/or situation.

A works by saying that there are people who responded "honest" who should actually be in the "dishonest" category, which distorts the survey results.

C introduces levels of dishonesty, which adds a new variable that wasn't considered in the original stimulus.