Tom claims that, because the data pairs (1, 4), (2, 6), (3, 9), and (4, 13.5) have y-values with a common ratio, the are best modeled by a quadratic function? What is his error?