Over the part two deca&s there has been a marked growth in the &d for dota for economic application in agriculture, due largely to an upsurge in the use of computers and computable modek, a growing awareness of the value of economic analysis and laming, and a continuing close institutional direction of the udutry. In adition, a greater commercialisation in forming with its concomitant pressures towar& the need for sound business organisation, has a d e d to the &mad for management data. However, reference to the literature in agricultural economics shows that, although there have been recent improvements, the supply of data of the required standud of reliability has not, on the whole, kept pace with the &mad.It is against this background that the review first considers the importance of data, their role in a decision-making fiamcwork and their desirable attributes: before turning to examine the reasons why an information gap may develop, and measures that mi ht be specifically on the ways in which rimary dotaexjurimental and non-experimentalare generatex together with comments on their utiliration, and improvements that might be maak Lastly, the manipulation of data is briefly reviewed.Although the mam emphasis in this article is placed on data for various applications in the management field, there is much that is pertinent to a wider sphere of interest. taken to ameliorate the problem. Attention is then focuse d more DATA IN AGRICULTURE 29 1 Figure I Information formatian l i m n mind inf wm at i o n Data __* formation t Problem c-Y Decision N m data f---Action -New problem -Source: McDonough (1963).Furthermore, when data are viewed in a decision-making context the areas mentioned in the title of this reviewmanagement research, policy and adviceappear not :as separate fields of endeavour, but as :closely related functions within an overall system. Nieto-Ostolaza (1973). for instance, conceives the process of problem-solving within an agrarian information-system, as being made up of a number of sets of interdependent functions (shown in the circles in Figure 2).* Data input is transformed into information and eventually into action by various 'processors', with the output of one becoming the input of another (moving from left to right), so enabling each to play its pertinent part within the whole system. For example, it might be the function of a data bank to collect and collate data; of research, to test hypotheses with the data and to elaborate models consistent with them; of advice, to simulate real situations using both the research models and information from the data bank (this might be regarded as 'applied' research as contrasted to 'fundamental research' in the previous stage); and of planning, to determine the goals and constraints relevant to specific situations. Feed-back (moving from right to left) is also important as, for instance, when the research function suggests what data ought to be collccttd, or when the advisory function suggests the need for improvements in models.It must be stressed that ...