Our findings provided further evidence for a simple negative conclusion: there is
no comprehensive and coherent ‘attitude order.’ This is not a message of despair.
The phrase ‘‘Individual I likesrdislikes to extent X the description D of object O,
considered in context C’’ is, at least in principle, subject to measurement, verification
and modeling. We already know, for example, that different measures of liking
will yield similar estimates of X, and that if two objects spontaneously evoke the
same context C, measurements of their relative preference by liking and by choice
will probably be consistent. Attitudes do not lack structure, but their structure is
vastly more complex than the structure that economic analysis attributes to human
preferences.