Standard saving theory - which assumes the «exponential» discounting of future utilities through to the end of the life cycle and refers, under an uncertain future, to the criterion of expected utility - uses only two preference parameters to explain asset behaviour: relative risk aversion and the discount rate. The clear mismatch between this theory's predictions and actual observation has given rise to the development of more realistic non-standard models - unexpected utility, hyperbolic discounting, etc. However, these models have to use more independent preference parameters to be able to agree with the laboratory and survey data: empirically, this proliferation runs into a dead end. We have therefore adopted a middle line using two «pivotal» preference parameters ­ one for risk and the other for time ­ whose definitions differ from the standard framework. The risk parameter characterises the general attitude to risk rather than actual aversion. Likewise, the long-run time preference is flanked by two other types of parameters measuring short-run impatience and altruism with regard to descendants. Correlatively, the empirical approach adopted proves to be both qualitative and eclectic. Aware of the disappointments of experiences to date in the English-speaking world, we propose purely ordinal preference measures, called «scores», which summarise the interviewee's responses to a multitude of all kinds of generally concrete questions ­ behaviour, opinions, intentions, etc. ­ covering a wide range of areas ­ consumption, health, work, financial management, family, retirement, etc.