So the next step, after you find an association, is to try to establish causation. For example a study could be done where volunteers are recruited and randomly assigned to bike or not bike to work for a set period, with pre- and post-cycling measures of happiness.