A nonparametric test of demand theory, determining the consistency of
price-expenditure data with the Generalized Axiom of Revealed Preferen
ce (GARP), is applied to household microeconomic data for the first ti
me. The fraction of consumption bundle comparisons that violate GARP i
s proposed as a summary statistic to indicate the consistency of a dat
a set with revealed preference axioms. This violation rate is calculat
ed among consumption pairs with similar expenditures to increase the p
ower of the test. The results suggest that similar household's consump
tion choices and budget constraints are generally consistent with the
joint hypotheses of optimizing behavior and common preferences.