Measuring Utility with Cardinal Numerically Weighted Point of View Question

I’m working on a philosophy writing question and need an explanation and answer to help me learn.

Questions:

1. According to Rosenberg (p.102), cardinal (i.e., numerically-weighted) measurability of utility relies on measuring subjective assessment of utility, and ordinal ranking of preference can avoid this requirement while arriving at a similar theoretical result. This is the benefit of using “preferences” over “marginal unitlity”, but what is the cost?

2. The theory of revealed preference (p.104-106) wants to shift the explanatory task of economics. “The starting point of economics is the consequence of, not the cause of, individual choice” (105). As such, theorists of revealed preference does not need to care about people’s reason for doing something. According to Rosenberg, What problem does this approach has?

3. On p.116, Rosenberg cites the study on the buying and selling of coffee mug. How does the result of this experiment shows a systematic departure from purely rational choice?