Published online by Cambridge University Press: 21 May 2010
This paper investigates the distribution of distance associated with marriage migration in the northern region of Rajshahi and the southern region of Khulna in Bangladesh. The study was conducted in 2007 on 2250 respondents who had migrated due to marriage. Of the wide variety of curves that fitted the distance–marriage/contact data, three are discussed: Pareto, exponential, and Pareto-exponential. Logistic regression models were used to identify the covariates of marriage distance migration. In general, the three functions work better for marriages, whereas Pareto-exponential functions are a superior fit for migrations and marriage distance. The models disclose that the distribution of distance is significantly associated with marriage migration (p<0.001). The Pareto-exponential model was 100% stable and its shrinkage was 0.000000125. The main covariates associated with short-distance marriage migration were respondent's education, father's education and religion, whereas age at the time of marriage did not play a significant role in marriage migration. The risk of short-distance migration was greater in higher- than lower-educated Muslim families.