The findings from two different types of statistical procedures are presented in this book. Firstly, differences in frequency between subgroups of children and families are described. These consist of univariate relationships between categorical variables. Examples include: the relationship between the presence or absence of medical problems and the frequency with which the child plays outside with friends; or, the relationship between social class and parental involvement in voluntary associations. All of these relationships were examined using chi-squared tests. Only those relationships that were significant at the five per cent level or less are described in the book.
Secondly, in order to examine the factors associated with different outcomes, we carried out a series of multivariate analyses. The results of these are described in Chapters 3, 4 and 5. The mothers’ scores on the Malaise Inventory indicating stress and depression, and the children’s scores on the BSQ indicating frequency and severity of behaviour problems formed two outcome measures. Other outcome measures were derived by combining categorical variables to form continuous indices. These measures include: child’s friends and play, sibling relationships and maternal satisfaction. The variables that were combined to form these indices are listed in Chapters 3, 4 and 5.
The analyses consisted of three stages. In the first stage, a series of univariate analyses of variance was carried out to examine the relationships between each child, parent and family characteristic as independent variables and the five outcome measures. The child, parent and family characteristics are listed in Table 3.11. With three exceptions, these have all been described in Chapter 2. The three exceptions are: family sharing, informal support and support from professionals. These variables were formed in the same way as the outcome measures, by combining all of the information from the interviews on these topics to form a continuous index. From these analyses, all relationships significant up to the ten per cent level of significance were tabulated.
The results of these univariate analyses were presented in terms