A scientist measures the efficiency of two catalysts. The likelihood ratio statistic $\Lambda$ for comparing a model with a common mean to a model with distinct means follows a $\chi^2_k$ distribution. If the scientist increases the number of constraints on the parameters, how is the degrees of freedom $k$ affected in the context of Wilks' Theorem?