Cross impact analysis

The term interaction analysis or English cross-impact analysis indicates a forecast technique that seeks correlations (English cross impact) represent between different future events that may occur, to analyze and consider the mutual impact. From experience we know that most of the events or developments in any way with other events and developments in relation marks (" correlation "). Many other forecasting techniques (such as the Delphi method ) can only view a delimited problem. The links individual events are not considered. This gap is filled by the cross-impact analysis. It will therefore, inter alia, used in the scenario technique.

The probability that a particular event is concluded, is influenced directly by the admission or non-occurrence of another event. The cross-impact analysis makes it possible to determine the probability of occurrence of an event depending on other events. It was developed in 1966 by Theodore Gordon and Olaf Helmer and resulted from the simple question: Can predictions are based on how to influence future events mutually exclusive? The first application of the cross-impact method was carried out in the context of a game ( "Future" ), the Gordon and Helmer developed for the Kaiser Aluminum and Chemical Company.

Method

1 Determination of the events: The first step of a cross-impact analysis is to find out the desired events. This step is critical to success. On the one hand all relevant can not recognized developments that nevertheless exert an influence, distort the result. On the other hand, a too detailed analysis that takes into account every possible event complicate the study unnecessary. As the number of interactions of the different pairs is equal to n (n = number of events ), one takes into account most of 10-40 events. This initial line-up of events is usually done by a summary of existing data and a survey of experts.

Second probability estimate: When estimating the probability of occurrence per event, each event is independent of / considered in isolation, ie without possible influences to take into account of other developments.

3 Conditional probabilities calculated: Then the conditional probabilities are determined. This process starts for each pair of events according to the following question: If event m occurs, what is the new probability of event n? This creates the cross-impact matrix:

Now the interpretation of the respective pairs of events can take place: Event 2 has a - viewed in isolation - probability of 0.70. But If an event 1, thus increasing the probability that event 2 occurs at 0.90 Similarly, the -. Considered in isolation - the probability that event occurs 3 0.35. But also enters Event 2, then the probability of occurrence of event 3 is reduced to 0.20.

4 Sensitivity analysis: Upon completion of the cross-impact matrix with a computer program performed several test runs, to better match the matrix. Here, events are selected at random, and computes the occurrence or non-occurrence and the thereby resulting influences on all events.

Assessment

A critical examination reveals some weaknesses of the cross-impact analysis:

• The selection and evaluation of the relevant factors is subjective
• The analysis is based on data pairs on - in the real world but can affect an event at the same time several developments
• The collection and analysis of data can be very time consuming - and so are some of the evaluation of a possible 30 events already 870 influences to calculate

But still is this very detailed study of different factors and their effects one of the biggest advantages of Cross Impact Analysis. You can give decisive impulses for alternative procedures or developing new ideas.