where vij can be interpreted as the contribution
of attribute j to the recreation value
of forest landscape i, and ai the interaction
effects between all attributes within forest
landscape i. While the overall recreation
value vi for each forest landscape iD.A/, (B),
(C), (D) was reported in Mattsson and Li
(1994a), in this paper we attempt to assess
the component values of and the interaction
effects between the different attributes (stand
types). This is done in two steps. First, the
component values of each attribute are estimated
as if they would dominate the whole
forest landscape i, and then the interaction
effects are determined by calibrating the estimated
model to fit the overall recreation
value for each forest landscape i.