Repeated migration : A simulation experiment based on Scottish materials

El-Rouby, Mohamed Gomaa (1976) Repeated migration : A simulation experiment based on Scottish materials. PhD thesis, University of Glasgow.

Abstract

A random sample of persons drawn from the Scottish Central Register was used to study the phenomenon of repeated migration in Scotland for the years 1939-1964. The existence of repeated migration flows in Scotland is established by the fact that a migrant, on the average, made approximately 2.315 moves compared with an overall average of 1.038 moves during that period. A study of the frequency of movement of the sample revealed that repeated migration is a characteristic of the single and young persons. Sex differential is of less importance than the age or marital status differentials. In general, changes in the life cycle from family formation to family dissolution are major stimuli for migration. The influence of past mobility experience in determining levels of future mobility, whether this past experience is expressed in terms of the frequency of previous movement or in terms of previous duration of residence, is investigated. The hypothesis that people who were more mobile in the past are more likely to move again in the future is tested and validated. These findings have considerable implications for the development of a general stochastic model of repeated migration. The model assumes that the propensity to migrate is not constant over time, but rather decreases as duration in the same place of residence increases, implying a linkage between previous mobility experience and future mobility prospects. The model also assumes that this propensity to migrate does not build up immediately after a move has occurred. There is probably some delay time during which the likelihood of a further move is negligible. The model is formulated in a simple mathematical form. It includes only three parameters defining respectively the general probability of moving, the delay time between a move and a subsequent move, and a constant term used to identify the mode of decline in the probability of moving in relation to the duration of residence since last move. The model is tested by means of a simulation technique. Hypothetical samples are simulated and a distribution of movement for each sample is generated. Such a distribution is compared with the actual distribution of movement and the Kolmogorov-Smirnov two sample test is used as a measure of goodness of fit of the model to the observed data. Simulation experimentation is a necessary media in order to minimize the value of Kolmogorov-Smirnov statistic to imply non- significant differences between the simulated and actual distributions of movement. Methods of locating optimum conditions operating upon a system which maximize or minimize features of it are utilized. These methods were developed by Box and associates for the study of chemical processes. According to these methods, the response surface is first represented by a plane whose coefficients are estimated using a suitably arranged experiment in the experimental region. The signs and magnitudes of the coefficients determine the direction of greatest gain in the response towards a near-stationary region. The response surface in this region is represented by a polynomial of higher degree than the first. The sig-ns and magnitudes of its coefficients determine the nature of the surface and consequently the minimum or mximum can be located using a few more extra points. These methods are adopted to determine the levels of model parameters which produce the best fit of the model to the observed data. Kolmogorov-Smirnov statistic is treated as the response to be minimized and the model parameters as the independent variables that yield this response.F our consecutive experiments comprising a total of 77 simulations are performed along the above lines. The final outcome of these experiments is a fall in the value of response to a level equal to 0.017 implying that the differences between the simulation and actual samples, with a high level of probability, are random variations and that the two samples are drawn from the same population. This indicates that the model efficiently describes the actual patterns of repeated migration in Scotland and may eventually be used as a general explanation of the repeated migration phenomenon. The effect of disaggregating the data by age and sex is assessed by fitting the model to selected age and sex subgroups. But certain constraints prevent such analysis from being thoroughly completed. However, the analysis suggests that sex and age variables as well as other relevant variables are important factors in formalizing a significant relationship between the probability of moving and the past history of movement.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Additional Information: Adviser: T E Hollingsworth
• Keywords: Demography
• Date of Award: 1976
• Depositing User: Enlighten Team
• Unique ID: glathesis:1976-72073
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 17 May 2019 13:07
• Last Modified: 17 May 2019 13:07
• URI: https://theses.gla.ac.uk/id/eprint/72073