Functional linear regression model where the responses are functions and the predictors are scalar vectors is considered. B-spline based functional F test (BF test) which is a hybrid of B-splines and functional F test is illustrated by an application from a robust design study. Simulations are conducted to compare the powers of BF test with functional F test and B-spline based multivariate likelihood ratio test (BM test) under three covariance structures. Robustness of Gaussian process is investigated as well. Results show that BF test and functional F test are more reliable than BM test with certain advantages and Gaussian distribution for assumed error process is not crucial.