Robin Hanson on Overcoming Bias recently posted links to data that religious people and people holding conservative political beliefs report being happier than heathen liberals. There was much debate over whether this should mean we should adopt conservative outlooks, and whether seeking truth was an inherent value even if it meant being less happy, but my problem is with the data: it is all based on self-report.
I am willing to go along with the idea that we have an a priori reason to believe a statement just because the statement is made up of language whose sense we can understand. So, if there were no further evidence, we would be justified in believing the self-reports of happiness. But, in this context, there is a custom of members of this group lying about this particular fact (happiness) - that is, the evangelical Christian custom of "witnessing," which entails acting happy and rich and perfect so that pagan nonbelievers may see how happy Christianity makes one. Conservative social movements such as Amway also encourage false displays of wealth and happiness. We have reason to question the sincerity of these self-reports of happiness.
But even if we allow the sincerity of these self-reports of happiness, we may still question their validity. It is extremely difficult to make an accurate overall evaluation of one's happiness. In fact, the tendency is to overstate one's happiness, a phenomenon known as the optimistic bias. Depressed people are far more accurate than non-depressed people in making predictions about the future based on evidence, a phenomenon known as depressive realism. Given the tendency toward unjustified optimism, we should doubt not only the self-reports of conservatives and the religious, but of all people who report happiness.
But, of course, how can we get at happiness if not through self-reports? Would it be ludicrous to try to develop some cheap and tasty qualia of happiness, or even qualia of suffering, which might be easier to measure? Wakefulness is often used as a proxy for consciousness where consciousness can't be reported (as with fetuses). Could there not be an objectively observable proxy for happiness that could be demonstrated to be so closely correlated with happiness that we'd trust it over a self-report?
As noted, suffering may be easier to measure objectively than happiness. Suicide rates may give an objectively verifiable estimate of the unhappiness in a given population. Do social conservatives kill themselves less frequently than liberals? However, given the importance of happiness both as an end, and of measuring happiness to figure out which policies might increase it, qualia of happiness seem to be worthy things to locate.
I am not necessarily defining happiness as temporary positive affect, although positive affect might be part of the calculation.