The results from analysis of data from the Bangor Longitudinal Study of Ageing (BLSA) is organised in order to test two main propositions:
1) Using the motives for relocation given by older people a classification of types of move will be developed. It is possible to assign respondents to these categories, from the relationships between combinations of observable factors. Probabilistic equations are devised that state the relationship of factors to each other and a set of theoretical statements are achieved that describe each category of migration for rural communities in Wales.
2) The typologies that have already been used to categorise older peoples moves (Litwak & Longino 1987, Wiseman 1980) are not entirely supported by empirical evidence from the BLSA. It is hypothesised that, a) the best fitting model has a
greater number of types of move than the three proposed by Litwak and Longino (1987) but fewer than the eight suggested by Wiseman (1980), but regardless of the types of moves identified, b) a comparison between those who entered residential care and those people remaining in the community, supports Litwak and Longino's (1987) assumptions that formal service provision is insufficient for people with major chronic disabilities to remain at home and its inadequacies play a part in institutionalisation.
Logistic regression is used to identify the factors that are most likely to predict who will move and who will stay put. Exploratory latent class analyses is used to develop a model of types of moves from the reasons that people gave for relocating. Confirmatory latent class analyses is used to determine the adequacy of Litwak and Longino's (1987) and Wiseman's (1980) models. Qualitative data from intensive interviews with respondents illustrate the quantitative findings. The conclusions are organised in two broad themes: methodological issues and implications for social policy.