Our model, assumed to reproduce the structure of the international system, has been constructed on the basis of statistical information available for the system's units, i.e. the nations included into our sample. This means that the structure has not been observed directly but through the unit's movements on the four status lines income per capiita (I), education (E), urbanization (U) and economic differentiation (LD) forming the structure. This is not at all an unusual procedure. However, when assessing the model's adequacy the following must be considered. Since the structure is not represented exactly by the aggregated behaviors of the units and since we do not make the assumption that the structure can itotally explain the behavior of the units, the question of the model's adequacy for reproducing the structure cannot be answered by measuring the degree to which the model explains national behavior. Taking this into consideration, the model is supposed only to maximize the degree of explanation that can be obtained from a knowledge of the nations' configurations of positions (p. 54). For ithe construction of the model a multivariable, non-linear fitting procedure of the least square type has been used, characterized by the fact that the multi-dimensional curvilinear regression line minimizes the squared distances to the points given by the available statistical information. The resulting functions (the regression equations) supposed to describe the structure are continuous. The deviations from the model, i.e. the behaviors deviating from the predictions of the model, are interpreted as due to factors being independent of the structure and having their roots in the unit itself, for example in the unit's culture. Conversely, if the behavior of a unit conforms to the Ibehavior predicted by the model it is interpreted as determined by the structure of the system. Tihe model assigns a single mobility path to each unit at each moment in time (p. 55). All variables are aggregated on the level of nations. The simple independent variables of the model are 1(1) income per capita (= Gross Domestic Product per capita) (I), (2) primary and secondary education in terms of enrollments as percentages of corresponding age groups (E), (3) urbanizaltion measured by the percentage share of the total population living in cities with 100 000 or more inhabitants (U), and (4) the differentiation of the labor force into the three sectors of agriculture, industry and services (LD = labor division). This last index assigns increasing weights to the three sectors in Ithe order presented (pp. 55-56). In addition to these simple independent variables, all linear ratios (a1/bl) and products al . b1) of pairs of the simple independent variables have been included. These complex independent variables are intended to express relationships, i.e. interconnections, between the simple variables. The dependent variables are operationalizations of a unit's mobility on the four status lines, i.e. the increments (or decrements) of the four simple variables (AI, AE, AU and ALD). The time periods for calculating the increments are 1950-1955, 1955-1960, and 19601965. The choice of these time periods has been determined by the availability of data. In the procedure applied, those groups of independent variables (simple and complex) have been selected which best predict the increments. The simulation of the recursive model included 53 nations representing all the major world areas and all levels of development. Again, this sample was determined by the availability of complete sets of data for each nation (p. 56).