The Scottish Government

« Previous | Contents | Next »

Listen

ANNEX A -MULTI-VARIATE ANALYSIS RESULTS

Description of the analysis

The fact that there is a relationship between key independent variables such as income levels, age of mother at birth of sample child, and level of mother's education, means that it is difficult to establish the key drivers of differences in the observations contained in simple bivariate analysis. For example, is the relationship between age of mother and use of grandparental support simply a function of the fact that younger mothers are more likely to live with grandparents? By using multivariate analysis (logistic regression) to look at the impact of a number of variables simultaneously on people's use of grandparental support, for example, we can find out whether the circumstances of younger mothers are distinct once other factors, such as living with a grandparent, are controlled. The results of these analyses are presented in the following tables.

All of the regression models included the following independent variables: parity, household income, mother's educational qualifications, mother's employment, respondent NS- SEC, age of mother at birth of sample child, area deprivation (whether or not family resided in an area within the 15% most deprived datazones) and area urban-rural classification (urban or rural according to a grouped version of the Scottish Executive 6-fold classification). Further, this technique requires that the outcome variable of interest has two-categories therefore for each model, each case was coded to reflect whether or not they fell into the category of interest.

The regression results are presented as odds ratios for each independent variable, all of which have a significance value and 95% confidence intervals attached. Odds ratios estimate the effect of each individual independent variable on the dependent variables, adjusted for all other independent variables in the model. Logistic regression compares the odds of a reference category (shown in the tables in brackets) with that of the other categories. An odds ratio of greater than one indicates that the group in question is more likely to have this view or demonstrate this characteristic than is the chosen reference category, an odds ratio of less than one means they are less likely. For example, Table A1 shows that the odds of people with no formal qualifications being amongst the group of people most likely to find it difficult to ask for help or advice, are twice as large as for people with qualifications at Higher grade or above.

As well as significance scores, odds ratios and confidence intervals, the regression tables display the results of two statistical tests carried out with the regression analysis which help to evaluate how well the models predicted the outcome variable - Nagelkerke's R ² and Hosmer and Lemeshow's Goodness of Fit test. Nagelkerke's R ² is most often quoted in logistic regression as a measure of strength of association ranging from 0 to 1. The closer the R ² value is to 1, the better the model is at accurately predicting the value of the outcome variable. A value closer to 0, suggests that there are important explanatory factors which are not included in the model. If the result of the Hosmer and Lemeshow Goodness of Fit test is not significant (p>0.05) the model's prediction of the outcome variable is not significantly different from the observed values of the outcome variable and the model is predicting the dependent variable well, or has 'good fit'. Further notes on the regression analysis are included in Appendix A.

0.1 Analysis of data from each of the Growing Up in Scotland cohorts must be undertaken separately (because together the cohorts do not represent a coherent or real population and results would be misleading). For the purposes of space and simplicity, all analysis in this report uses only data collected from natural mothers in the birth cohort. Larger numbers in the birth cohort also allow more detailed analysis of the selected sub-groups.

Results

Table A1 Logitistic regression model detailing factors related to having most difficulty seeking help or support: birth cohort

Table A2 Logitistic regression model detailing factors related to low access to sources of informal support: birth cohort

Table A3 Logitistic regression model detailing factors related to low use of sources of informal support: birth cohort

Table A4 Logitistic regression model detailing factors related to receiving a low-level of support from the child's grandparents: birth cohort

« Previous | Contents | Next »