We use two waves of a population based survey (the RAND American Life Panel) to investigate the relations between various evaluative and experienced well-being measures based on the English Longitudinal Study of Aging, the Gallup Wellbeing Index, and a 12-item hedonic well-being module of the Health and Retirement Study. In a randomized set-up we administered several versions of the survey with different response scales. Using factor analysis, we find that all evaluative measures load on the same factor, but the positive and negative experienced affect measures load on different factors. We find evidence of an effect of response scales on both the estimated number of underlying factors and their relations with demographics. We conclude that finer response scales allowing more nuanced answers offer more reliability. The relation of evaluative and experienced measures with demographics are very different; perhaps the most striking aspect is the lack of a consistent relation of experienced well-being measures with income, while evaluative well-being is strongly positively related with income.