This thesis deals with cross impact analysis for technology assessment. The focus of the thesis is the technique of cross impact matrix.
Critical literature review on the cross impact analysis is given with respect to event cross impact analysis, trend impact analysis, and scenario generation method from the early days of development in late 1960's. Of these techniques SMIC 74 which has mathematical rationale of consistent probability is explained in detaily, including its contents, procedural variations and critiques.
In this study a new model of cross impact analysis using goal programming based on the scenario generation technique is developed. In the model, cross impact is fully considered using a priority factor of goal programming.
Deviational variances of probability judgement are minimized according to priority factors. Specifically, the priority factors are $P_1$, $P_2$(priority 1, priority 2). $P_1$ is given to the deviational variables of conditional probabilities which considers cross impacts, $P_2$ to those of independent future probabilities. In each subgoal weighting values are also used to give better reliability of probability estimations. The objective function in the model is given as follows :
$Min P_1 \displaystyle\sum\sum_{j>i}^{n n-1} (w_{1i}^+d_{ij}^++w_{1i}^-d_{ij}^-)+p_2\sum_{i=1}^n (w_{2i}^+ d_i^++w_{2i}^-d_i^-)$ This formulation covers the optional character of objective function of SMIC 74 i. e., Max (Max $π_k$). The model reduces drastically the amount of computer runs.
This new technique is applied to the assessment of the air pollution in Seoul Metropolitan area in 1990 and 2000. The results of analysis give us the following findings :
1) Cross impact analysis using goal programming has more meaningful solutions as comparing to the results by SMIC 74, simulation.
2) Theoretical rationale of the objective function in the newly developed technique is more appropriate than that of SMIC 74.
Further research should be done to improve the technique related to weighting problem.