Although I would gamble Schwartz has more ideological than empirical reasons for his conclusions, is there trend data available over multiple years? I don't think a point estimate is any way to gauge something like "happiness with respect to time." As for surveys, from what I know of the experimental literature, economists seem to put very little credence in surveys aside from gathering rather concrete data (e.g. demographics, independently verifiable data). My impression is that survey research is generally looked down upon -- if for no other reason than similar research in other fields is often highly contestable and unsophisticated (in both senses). Surveys also beg the question of the incentive and ability of respondents to answer in an externally valid way. Monetary incentives used to help ensure externally valid behavior seem to be a publishing requirement of modern experiments -- incentives completely lacking by nature in a survey. However, this may be completely different in particular subfields.
-- John Morrow
