Embedding, the notion that respondents to contingent valuation (CV) qu
estions often value more than the researcher intends, has engendered e
xtreme views. These range from the suggestion that embedding is so sev
ere that it renders CV useless to the assertion that embedding can be
eliminated by providing sufficient information in a survey. This paper
examines three alternative explanations for embedding: 1) the purchas
e of moral satisfaction, 2) independent valuation and summation, and 3
) mental models of joint products. Several studies that shed light on
the nature of the embedding problem are presented and we examine wheth
er it is possible to test for embedding through follow up questions th
at obtain self reports from respondents. We show that answers to these
debriefing questions predict whether or not different groups of respo
ndents pass a scope test (i.e., a statistically significant difference
between groups valuing different levels of the same commodity). We re
ject the view that increased market context can solve the embedding pr
oblem. Rather, embedding may depend on the nature of the commodity its
elf. (C) 1998 Elsevier Science B.V.