recently published works, Michael (1972, 1973) exhibited evidence that offers tentative support for the assertion that education not only affects consumption indirectly by raising productivity at work, but that in addition, education directly affects the consumer's lifetime real income (utility) stream. In Michael's language, education increases nonmarket productivity the efficiency with which the consumer combines market goods and leisure time to generate utility. To obtain his empirical results, Michael assumed that education enters as a parameter in the utility-production functions which characterize the consumption of goods and leisure time. He further assumed that the shifts in these functions caused by education are Hicks' neutral in the sense that both the ratio of marginal products of time and market goods, and the ratio of marginal utilities for any two of the more basic goods produced by combining market goods and time, are independent of attained educational level. These assumptions allowed Michael to conclude that as the consumer's educational level rises, the composition of market purchases will shift toward goods whose income elasticity is greater than one, and away from goods with less than unitary elasticity. A cross-section analysis of consumer expenditures showed that this hypothesis could not be rejected. As Michael himself indicated, the derivation of his test is complex; there is thus a good possibility that a quite different theory could have the same empirical implication. For example, education could affect choices by changing the goods that the consumer considers important in his or her consumption bundle.' This effect would show up as an alteration in the composition of market purchases; it could be quantified as education elasticities for a number of consumer expenditure categories; and the relationships between these elasticities and income elasticities could be just as Michael discovered in his work. Inasmuch as educational attainment tends to be highly correlated with income, other factors not being held constant, it should not be surprising to find that education and income have similar effects on the composition of market purchases. On the other hand, by restricting attention to the choice of spending income on a composite consumption good, or foregoing income to enjoy leisure, it is possible to derive some simpler tests of the direct effects of education on choices. Further, if one is willing to identify changes in the shape of indifference curves as the preference structure effect of education, then nonmarket productivity effects can be distinguished from such lifestyle effects.