Well, they didn't actually show you how they used Ek scores to predict agreeableness. They simply stated that they did, which should be enough to rule out answer C. What Figure 2 shows is their actual scores plotted against their predicted scores. If Ek scores entirely predicted agreeableness, you would expect the slope of the line cross through zero and have a slope of exactly one. In other words, y = x, or actual score = predicted score. In this case, it looks like Ek is a rather poor predictor.
