02.06.2014 Views

Evaluation of the Australian Wage Subsidy Special Youth ...

Evaluation of the Australian Wage Subsidy Special Youth ...

Evaluation of the Australian Wage Subsidy Special Youth ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

156<br />

natural attrition have missing values for <strong>the</strong>se cases, as <strong>the</strong>y are defined in <strong>the</strong> 1986<br />

survey. Accordingly, in <strong>the</strong> table <strong>the</strong>y are not presented since <strong>the</strong> overall mean would not<br />

be representative <strong>of</strong> <strong>the</strong> same observations as <strong>the</strong> rest <strong>of</strong> <strong>the</strong> table. Effects upon <strong>the</strong>se<br />

variables are dealt with in <strong>the</strong> later section where <strong>the</strong> source <strong>of</strong> data reduction is<br />

controlled for.<br />

Several characteristics in Table 5.2 exhibit statistically significant differences between<br />

<strong>the</strong> final sample and those lost from <strong>the</strong> sample: whe<strong>the</strong>r <strong>the</strong> partner was employed,<br />

schooling to year 9 and year 11, <strong>the</strong> length <strong>of</strong> <strong>the</strong> last job held, work limiting health<br />

problems, state <strong>of</strong> residence at time <strong>of</strong> interview, and <strong>the</strong> rural-urban nature <strong>of</strong> <strong>the</strong><br />

location. The SYETP treatment variable also shows evidence that it was affected by<br />

sample reduction. Thus both outcomes and regressors appear affected in <strong>the</strong> univariate<br />

analysis.<br />

It is interesting to examine how those lost from <strong>the</strong> sample compare to those remaining.<br />

Past studies, such as Hausman and Wise (1979) found patterns indicating particular strata<br />

<strong>of</strong> <strong>the</strong>ir dependent variable were not being observed, in <strong>the</strong>ir case parts <strong>of</strong> <strong>the</strong> earnings<br />

distribution were truncated, and certain subgroups more likely not to be surveyed.<br />

Fitzgerald et al. (1998a) also found lower socio-economic groups were more likely to be<br />

affected. Here, amongst those lost from <strong>the</strong> sample, when compared to those remaining,<br />

fewer had employed spouses in 1984, fewer had obtained only year 11 at school while<br />

more had year 9 <strong>of</strong> school, and fewer had health problems that affected <strong>the</strong>ir work. The<br />

length <strong>of</strong> <strong>the</strong> last job held differed, for more had held jobs <strong>of</strong> 1-2 years while less had<br />

held jobs <strong>of</strong> 2-3 years. Of those lost, fewer were first interviewed in South Australia or<br />

Nor<strong>the</strong>rn Territory and more in New South Wales or <strong>the</strong> <strong>Australian</strong> Capital Territory. The<br />

location <strong>of</strong> <strong>the</strong> interview was important, as amongst those lost fewer came from rural<br />

areas or country towns, and more were from cities. Fewer <strong>of</strong> those lost from <strong>the</strong> sample<br />

had taken part in SYETP.<br />

There is no clear picture <strong>of</strong> those lost being from lower socioeconomic groups or having<br />

a more disadvantaged labour market position. While <strong>the</strong> educational capital is slightly

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!