A restaurant owner wanted to gather some data on customer taste preferences about two new salads on the menu, one a mixed-greens salad, and one a roasted vegetable salad. A random sample of diners was selected, and each diner was asked to taste both of the new salads and indicate which was preferred. There were 35 diners in the sample. Of these, three liked the salads equally. Of the remainder, 20 gave a higher rating to the mixed-greens salad. Is there evidence, at the 3% level of significance, that diners prefer the mixed-greens salad?