Evaluation of the Australian Wage Subsidy Special Youth ...
Evaluation of the Australian Wage Subsidy Special Youth ... Evaluation of the Australian Wage Subsidy Special Youth ...
168 been higher, and the location of the interview differed to those remaining with fewer from South Australia and the Northern Territory, Western Australia and Tasmania, more from cities and fewer from country towns and rural areas. The group of observations lost due to analytical selection had different features from both those remaining and from those lost by natural attrition. For those lost for analytical reasons, the means of more characteristics were affected than for those lost through natural attrition. Compared to those remaining, fewer of those lost were married, had employed spouses, apprenticeships, year 10 of school, mothers in the managerial occupations, or had held a job for 3 years or more, while on the other hand more had work limiting health problems and year 12 of school. The key outcome variable, employment in 1986, was significantly lower for those lost by analytical selection. The effects of analytical selection reinforced those of natural attrition for three variables, SYETP treatment, South Australia and the Northern Territory and rural areas. For each of these aspects, fewer cases were observed for those lost by natural attrition. The process of analytical selection worked in the same direction as that of natural attrition, and so of those lost, fewer had taken part in SYETP, fewer were from South Australia and the Northern Territory and fewer were from rural areas. There is a weak suggestion that among those lost due to natural attrition, the average proportion of unemployment in 1984 is lower, and of those dropped by analytical selection fewer were observed in employment in 1986. However, it is considered that this is not sufficient evidence to support the hypothesis that key subgroups are disproportionately affected in terms of the dependent variable labour market status. This question is readdressed in the multivariate analysis that follows. Drawing together the findings of the univariate analysis of sample reduction, it can be said that profiles of some important regressor variables, and the key treatment variable SYETP, show evidence of being statistically different after sample reduction. Segregation by the two types of sample reduction, shows natural attrition and analytical selection are
169 both sources that impose changes in the profiles, and they have differing impacts. However, univariate analysis cannot account for the correlations between these variables and so the t statistics do not represent independent effects. As such, it is maintained that there is as yet no strong indication of serious problems arising from sample reduction. Further tests accounting for effects of sample reduction in a multivariate context are now applied. As the pattern of differences varies by the form of sample reduction being natural attrition or analytical selection, it seems that it will be important to maintain the breakdown of sample reduction by source in further examinations. Accordingly, the modelling of sample reduction, and any weights arising, takes separate account of natural attrition and analytical selection.
- Page 133 and 134: 117 covariates that influence the a
- Page 135 and 136: 119 (7) E(Y c | D=1) = E P(X) {E[Y
- Page 137 and 138: 121 For CIA to be plausible, a ‘r
- Page 139 and 140: 123 employment and programme partic
- Page 141 and 142: Highest qualification in 1984 (1.56
- Page 143 and 144: 127 4.6 Distribution of the propens
- Page 145 and 146: 129 Figure 4.3 Histograms of estima
- Page 147 and 148: 131 Table 4.5 Summary statistics fo
- Page 149 and 150: 133 Table 4.5, that the variance of
- Page 151 and 152: 135 Table 4.6 Matching results, Sin
- Page 153 and 154: 137 Table 6.3 using Swedish data wi
- Page 155 and 156: 139 matching is the ability to weed
- Page 157 and 158: 141 Table 4.7 Matching results, All
- Page 159 and 160: 143 the unobserved component. If th
- Page 161 and 162: 145 5: Study 3 Attrition and non-re
- Page 163 and 164: 147 occur by design, because the mi
- Page 165 and 166: 149 (1990) extended and improved th
- Page 167 and 168: 151 (10) A* = δ 0 + δ 1 x +δ 2 z
- Page 169 and 170: 153 again from September to Novembe
- Page 171 and 172: 155 5.5.2 Univariate examination of
- Page 173 and 174: 157 lower, the job lengths are only
- Page 175 and 176: 159 Work limited by health 1984 0.1
- Page 177 and 178: 161 The characteristics of the SYET
- Page 179 and 180: 163 para-professional Mother not em
- Page 181 and 182: 165 comparison group where the shar
- Page 183: 167 5.5.4 Attrition: natural attrit
- Page 187 and 188: 171 para-professional Father not em
- Page 189 and 190: 173 work in later sections, this su
- Page 191 and 192: 175 Table 5.6: Effect of selection/
- Page 193 and 194: 177 appropriate to discard these fr
- Page 195 and 196: 179 Australia/Tasmania. Amongst tho
- Page 197 and 198: 181 Table 5.5a Summary statistics b
- Page 199 and 200: 183 5.6.1.2 Effects of the non-resp
- Page 201 and 202: 185 3 years + -0.35 -0.47 -0.34 -0.
- Page 203 and 204: 187 5.7 Multivariate analysis of ef
- Page 205 and 206: 189 proportion of time spent unempl
- Page 207 and 208: 191 post-school qualification, and
- Page 209 and 210: 193 Generally, those variables foun
- Page 211 and 212: 195 longj0 Longest job by 1984 < 1
- Page 213 and 214: 197 adopted in order to maintain co
- Page 215 and 216: 199 6: Study 4 Weighting to counter
- Page 217 and 218: 201 Table 6.1, part A Employment eq
- Page 219 and 220: 203 Methodist 0.133 0.261 (0.77) (1
- Page 221 and 222: 205 CEP referrals 1984 0.143* 0.128
- Page 223 and 224: 207 6.2 Results of weighting the PS
- Page 225 and 226: 209 The distribution of the propens
- Page 227 and 228: 211 Table 6.3 Weighted probit used
- Page 229 and 230: 213 (0.76) Tradesperson mtrad 0.20
- Page 231 and 232: 215 Table 6.5 Summary statistics fo
- Page 233 and 234: 217 Table 6.7 Matching results, sin
169<br />
both sources that impose changes in <strong>the</strong> pr<strong>of</strong>iles, and <strong>the</strong>y have differing impacts.<br />
However, univariate analysis cannot account for <strong>the</strong> correlations between <strong>the</strong>se variables<br />
and so <strong>the</strong> t statistics do not represent independent effects. As such, it is maintained that<br />
<strong>the</strong>re is as yet no strong indication <strong>of</strong> serious problems arising from sample reduction.<br />
Fur<strong>the</strong>r tests accounting for effects <strong>of</strong> sample reduction in a multivariate context are now<br />
applied. As <strong>the</strong> pattern <strong>of</strong> differences varies by <strong>the</strong> form <strong>of</strong> sample reduction being<br />
natural attrition or analytical selection, it seems that it will be important to maintain <strong>the</strong><br />
breakdown <strong>of</strong> sample reduction by source in fur<strong>the</strong>r examinations. Accordingly, <strong>the</strong><br />
modelling <strong>of</strong> sample reduction, and any weights arising, takes separate account <strong>of</strong> natural<br />
attrition and analytical selection.