ABSTRACT
This thesis is based on data made availble by the World Health Organisation. The data base consists of dailt menstrual bleeding records of 4817 women using one of ten fifferent fertillty regulating methods. After introductory chapters that discuss the data base and review previous work on the analysis of menstrual bleeding patterns. two new approaches to the analysis of data are consideres in detail. An approach based on techniques used in the analysis of time series data is discussed in Chapter 4. The major part of the thesis, Chapter 5 to 8, describe the use of generalised linear models to model the lengths of bleeding and non-bleeding events. A time series approach to the analysis of menstrual bleeding data has been introduced by pochobradsky (1970). In this thesis his approach is extended by using principal compounent analysis to analysis the periodicity of vaginal bleeding patterns of women., Statistics that summarise the shape of the periodogram (a powerful tool in the study of rhythmic processes) of daily bleeding data are used to describe the vatiation in the bleeding patterns among different contraceptive ueers and to summarise the bleeding experience of women under differet contraceptive methods. In modelling menstrual bleeding data. interest is centered on the relationship between a woman's bleeding....