The studies of migration are classified into the analyses of migration factors and the investigations of spatial movements of migrants. In the former study, known as push-pull theory, it has been discussed that migration occurs as a movement to keep the regional equilibrium of income level and employment condition. On the other hand, the study of the latter has been developed in connection with the gravity model in which many problems have appeared. The value of distance exponent n seems to be variable due to various sorts of movements, varying urban size of the destination and in connection with service areas of central places, while functional distance may be actually more useful than physical distance like Stouffer's intervening opportunity hypothesis. The mobility difference of inhabitants as a mass also should be taken into consideration. Though the migration model is Improved by considering these factors, another direction in the study also exists in its improvement where better understanding of migration movements is expected by adding other migration factors to the gravity model as G. Olsson and R.A.Hart have done. The author also attempts to find out a better explanation of migration intensity with multiple regression model by adding other migration factors to the distance which has been measured by road distance. As a case study the author selected Hiroshima prefecture and researched into the migration between Hiroshima city and other communities in the Prefecture. The correlation coefficient between gross migration intensity and distance in Fig. 2, ( r = -0.757, n =106) is considerably low as compared with its values of visiting patients ( r = -0.875), commuters (r = -0.872) and driving learners ( r = -0.871), though value of inpatients (r = -0.765) is similar to value of gross migration intensity. It suggests that gross migration intensity such as the case of inpatients to Hiroshima city is influenced by many other factors except distance factor. As shown in Fig. 2, the distribution of residuals between the actual values of gross migration intensity and the estimated values from gravity model in the formula (1) mainly indicates the active values in the northern part of the prefecture on the one hand, and the negative values in the coastal region on the other hand. The negative values are especially large in the southeastern part which is not included in the migration area of Hiroshima city. In the construction of multiple regression model the author has introduced the following migration factors X_2-X_7 into the formula. 1 ) Areal difference of income level should have effect on volume and direction of migration movement. Consequently the author took up per capita income in i-community/per capita income in Hiroshima city (1967) as X_2 factor and the employee ratio of primary industry (1970) as X_3 factor. The X_3 factor is often used as an alternative one, because it sharply indicates areal difference of living standard. 2) The commuting ratio from i-community to Hiroshima city (X_4) may be conversely correlated in job opportunity in i-community and its surrounding area. 3) Central place system affects the migration movement. If the distance is equal, few migrants will move to the service area of another central place. Therefore, gross migration ratio to Fukuyama, Onomichi, Mihara and Fuchu cities from i-community was considered as X_5 factor. 4) When i-community increases gross migration ratio to six metropolitan prefectures (X_6), it may decrease the contact with Hiroshima city. 5) Though the ratio of net migration from i-community to Hiroshima city (X_7) should be theoretically indifferent from the ratio of gross migration, the latter may become larger, the larger the former is.
 View full abstract