Passions, People and Appreciation: Making Volunteering Work for Young People 

Notes to Tables

1. All tables (except Tables 1 and 4) are based on data from the 1995 panel of the Longitudinal Surveys of Australian Youth. Further information on the survey can be obtained at http://www.acer.edu.au/ through the research and then vocational education tabs.
2. Throughout the tables based on LSAY data, missing values in any of the background variables were imputed using a variant of hot-decking.
3. Throughout the tables based on LSAY data, percentages are weighted but any values for the number of cases on which an estimate is based are unweighted. It is the unweighted values that better convey a sense of the evidence on which an estimate is based. Inconsistencies may result if weighted estimates are combined using unweighted numbers.
4. Statistical significance corresponds to probability values (presented in italics or parentheses) of 0.05 or less.
5. In Tables 3, 8, A6, A7 and A8 a number of separately identified categories listed under "other community volunteer work" were recoded to the main prompts in the question because they appeared identical to the main prompts.
6. The percentages responding to the various kinds of volunteer work listed in Tables 2 and 3 (and corresponding Tables 5, 8, A1, A2, A6 and A7) may sum to more than the total because respondents can participate in more than one type of volunteer work.
7. Items presented in square brackets in Tables 2 and 3 were not included in the interview questions but are derived directly from responses to those questions.
8. In Tables 2 and 3, na indicates not applicable. Since individual responses under the heading Anything else I haven't mentioned were unprompted, it was not possible to respond No.
9. In Table 5 (and corresponding Tables A1, A2 and A3) and Table 8 (and corresponding Tables A6, A7 and A8) the values in italics indicate statistical significance. They indicate the probability of no relationship between the variable and participation in the type of volunteer work. The values are calculated using weighted least squares.
10. In Tables 6 (and corresponding Tables A4 and A5) and Table 9 (and corresponding Tables A9 and A10) the value in italics next to each variable name indicate statistical significance. They indicate the probability of no relationship between the variable and participation in volunteer work at least once a month. They are calculated using weighted least squares.
11. Tables 7 and 10 report results from logistic regression. Logistic regression is an appropriate multivariate statistical technique for analysing dichotomous dependent variables. In this case the dichotomous dependent variable consists of the two categories participated in volunteer work at least once a month and participated in volunteer work less than once a month.

Independent variables are fitted to the model using a set of dummy variables – one for each category, except that one category must always be omitted as the base or reference category. The omitted category for each variable is indicated by four dashes.

The variable Respondent's country of birth, which was included in other tables, has been excluded from these multivariate analyses because it is too closely related to Parent's country of birth.

The values in italics in parentheses indicate the statistical significance of the entire variable in the model. It is derived from a log-likelihood chi-square which in turn is the difference between the log-likelihood chi-squares of the model containing the variable and not containing the variable. The p values also indicate statistical significance of each category of a variable compared to the omitted category. They are Wald chi-squares. The discussion in the text uses the log-likelihood chi-square for the entire variable.

The values in the columns labelled % are the linear estimates derived from the non-linear logistic regression by iteratively estimating the percentages implicit in the corresponding logits and constraining these percentages to be the weighted sum of the overall mean. These values are more easily understood than logits (or the corresponding odds ratios) and hence the discussion focuses on these columns.

Model 2 is a single model that simultaneously fits all the variables listed in Tables 7 and 10. The values listed under Model 1 are derived from a series of separate models that progressively add variables based on the order provided in Table 4. The values for Sex to Location are derived from a single model that fits these variables (listed under Background in Table 4). The values for School sector and Literacy & numeracy are derived from models containing respectively School sector plus the background variables and Literacy & numeracy plus the background variables. The values for Self-concept come from a model containing all the variables listed above it in the table. Similarly the values for Year left school and Current activity are derived from models containing that variable and all the variables above it in the table.

Results for Models 1 and 2 are presented to provide more information on the structure of the relationships that may affect any relationship between a particular variable and the dependent variable.