A goal of nuclear theory is to make quantitative predictions of low-energy nuclear observables starting from accurate internucleon forces. This work focuses on two approaches for calculating nuclear observables, the consistent application of ab initio methods for light nuclei and a systematic investigation of transition strengths in the p- and sd-shells. A major element of ab initio calculations is applying unitary transformations to soften the nuclear Hamiltonian and hence accelerate the convergence as a function of model space size. Here we study the evolution of the r² and electric dipole operator in the framework of the similarity renormalization group (SRG) method. The effects of the evolved operators, although seemingly small, are comparable in magnitude to the correction produced by including the chiral three-nucleon force. For my second investigation I turn to the p- and sd-shells to systematically compute E1, E2, M1 and Gamow-Teller transitions for several nuclei. My primary goal is to test the Brink-Axel hypothesis by computing strength functions starting from highly excited states. I find that although there are large fluctuations about the average strength, particularly at low excitation energy, there are consistent average giant resonances as initial excitation energy increases.