Conditions of Happiness, Kluwer Academic, 1984, Dordrecht/Boston.
Happiness can be assessed only by asking people about it. That is at least true for
'overall happiness' and 'contentment'. 'Hedonic level' can to some extent be inferred from
Several doubts are being raised about the quality of responses to questions on happiness: especially about the validity of direct questions on overall happiness. It is suggested that people do not know, that they are reluctant to discuss the matter, that they fool themselves, that they try to appear happier than they know they are, etc. Most of these doubts can be discarded on the basis of empirical evidence. It was for example shown that most people have quite definite ideas on whether they are happy or not and it is hence unlikely that questions on the matter tap hot air only. Not all objections can be discarded however; especially not the objection that people may fool themselves as well as interviewers by pretending to be more happy than they in fact are. Yet none of these doubts have been proven true either. It is as yet plainly unclear to what extent they apply. It has neither been ascertained whether any further, as yet unrecognized, distortions are involved. Unfortunately the matter can hardly be settled by testing for congruent' or 'concurrent' validity.
Next to doubts about validity there are questions about the technical 'reliability' of self-reports of happiness. It is objected that responses tend to be heavily biased by for instance interviewer characteristics, answer formats and contextual cues. Checked empirically, these distortions do not appear too dramatic, however.
Though not convincingly demonstrated, the various objections are still serious enough to be taken into account. They suggest at least four working rules: Firstly, self-ratings are to be preferred to ratings by others. Secondly,anonymous questionnaires work better than personal interviews. Thirdly, the context of the questionnaire as well as the key-questions must be focused clearly on the 'overall' appreciation of 'life-as-a-whole'. Fourthly, questions must leave room for 'no answer' or 'don't know' responses.
Fewer solutions seem available for the problem of comparison. We are not sure whether two people, both claiming to be happy, are in fact talking about identical levels of appreciation. This implies that respondents can be ranked for happiness only rather crudely. In practice this means that statistical correlations of measured happiness will be somewhat less pronounced than correlations of true happiness.
Possibly one or more of the various objections to happiness testing will in due course be convincingly substantiated; for the time being attempts to measure happiness deserve the benefit of the doubt.