【正文】 has been developed by economists and statisticians and is derived from decision theory. The basic premise of the theory is that we always have some preliminary knowledge about the occurrence of events that are relevant to our decisions. Additional information might modify our view of the occurrence probabilities and consequently change our decision and the expected payoff from the decision. The value of additional information is, hence, the difference in expected payoff obtained by reduced uncertainty about the future event. Information supports decisions, decisions trigger actions, and actions affect the achievements or performance of the organization. If we can measure the differences in performance, we can trace the impact of information, provided that the measurements are carefully performed, the relationships among variables are well defined, and possible effects of irrelevant factors are isolated. The measured difference in performance due to informational factors is called the realistic value or revealed value of information. For most information systems, particularly those supporting middle and top management, the resulting decisions often relate to events that are not strictly defined and involve probabilities that cannot be quantified. The decisionmaking process often is obscure and the outes are scaled by multiple and inparable dimensions. In such cases, we may either attempt to perform a multiattribute analysis or derive an overall subjective value. The subjective value reflects people39。 Singh, 1987). Need for automation An automated MIS system contains data just as a manual system does. It receives input, processes input, and delivers the processed input as output. Some input devices allow direct humanmachine munication, while others require data to be recorded on an input medium such as a magizable material (specially coated plastic flexible or floppy disks and magic tapes). The keyboard of a workstation connected directly to a puter is an example of a direct input device. Use of automation makes it possible to store immense quantities of information, to avoid many of the errors that find their way into manual records, and to make calculations and parisons that would be practically impossible in a manual system. Organization of a database Data are usually generated at the field level through transactionprocessing systems, but once the data are captured, any echelon along the organizational hierarchy may use them, provided that information requirements have been well defined, appropriate programmers have been implemented, and a means has been arranged for the sharing of the data. This would imply that the same data can be used by different sets of programmers。s administrative machinery is divided into districts, districts into subdivisions, subdivisions into blocks. A block is a group of villages and the basic unit for the administration of an agricultural extension programmer. Data collected at the block level need to be integrated at higher administrative levels to provide an integrated view at the district and state levels to support planning, monitoring, and decision making. However, the actual design may vary with the size of the state and other considerations. An integrated database for the entire state may be supported by a mainframe/miniputer at the state headquarters. Suitable programmes for the analysis of data may be designed to provide an interactive decision support system at the state level. Each district and subdivision may be provided with a mini/micro puter, depending on the volume of data to be handled. The puters in the districts and subdivisions may be worked with the state puter. The local data may be stored and processed in the district/subdivision, and the shared data with appropriate level of aggregation may be transmitted to the state headquarters to update the integrated database. The districts and subdivisions would have direct access to the integrated database with proper authorizations assigned to them through their passwords. The blocks may have only the inputoutput terminals connected to the subdivision puter to feed data t