NURSES’ RETENTION AND HOSPITAL CHARACTERISTICS IN NEW SOUTH WALES
APPENDIX 1
In this section of the paper we compare the analysis sample of RNs used above to the full sample of
RNs working in the NSW public sector.
Appendix Table 1: Sample means
|
VARIABLE |
FULL SAMPLE |
ANALYSIS SAMPLE | ||
|
MEAN |
STD. DEV |
MEAN |
STD. DEV | |
|
working 97 |
0.7894 |
0.4214 |
0.8050 |
0.4085 |
|
male |
0.0853 |
0.2794 |
0.0814 |
0.2734 |
|
age |
39.0865 |
9.6968 |
38.9509 |
9.4791 |
|
foreign born |
0.1425 |
0.3496 |
0.1266 |
0.3325 |
|
UK born |
0.0869 |
0.2818 |
0.0818 |
0.2741 |
|
NZ born |
0.0195 |
0.1382 |
0.0191 |
0.1369 |
|
foreign citizen |
0.0706 |
0.2562 |
0.0685 |
0.2526 |
|
not permanent resident |
0.0009 |
0.0306 |
0.0005 |
0.0221 |
|
work hours |
36.1319 |
15.9408 |
36.1520 |
16.0382 |
|
post basic qualifications |
0.5723 |
0.4876 |
0.5836 |
0.4887 |
|
years registered |
15.1886 |
9.0748 |
15.2330 |
8.8935 |
|
unemployment rate |
6.8348 |
3.0944 |
6.7019 |
3.3802 |
|
activity surgical |
0.2503 |
0.4332 |
0.2693 |
0.4436 |
|
activity psych |
0.1623 |
0.3688 |
0.1435 |
0.3506 |
|
job clinical |
0.2711 |
0.4446 |
0.2872 |
0.4525 |
|
job managerial |
0.0547 |
0.2274 |
0.0569 |
0.2317 |
|
region other metropolitan |
0.1071 |
0.3092 |
0.1079 |
0.3102 |
|
region non metropolitan |
0.1438 |
0.3509 |
0.2144 |
0.4104 |
|
sample size |
25568 |
16393 | ||
A comparison of the means for the two samples reveals relatively small differences in almost all
explanatory variables. The largest discrepancy is in the regional dummies: our analysis sample has
more observations in non metropolitan areas. Nurses working in metropolitan areas (including
capital cities) were more likely to omit the job postcode and in some cases, we could not match the
reported postcode to a specific hospital. Appendix Table 2 presents the coefficients from a probit
regression similar to that presented in left hand columns of Table 4 but run on the of all RNs working
in public hospitals in NSW.
A comparison of the coefficients of Appendix Table 2 with Table 4 shows that the effects of the
personal and job characteristics are very similar for the full sample and our analysis sample. The
only substantial difference involves the foreign born dummy variable which has a stronger positive
effect in the full sample.
A sample selection procedure incorporating the observations with missing hospital information is
problematic because of the lack of an instrument identifying the retention probability model from the
selection process. As a simple check on the possible selection problems we estimated a probit for
selection into the analysis sample and included the predicted probability in the retention probit. The
first stage probit included all of the variables listed in Appendix Table 2. In the second stage probit,
the results for the hospital variables were unaffected by the inclusion of the predicted probability and
the coefficient on this probability was insignificant.
Given these results, it is unlikely that such a procedure would add anything to the analysis presented
in the main text.
18
More intriguing information
1. MATHEMATICS AS AN EXACT AND PRECISE LANGUAGE OF NATURE2. The name is absent
3. ESTIMATION OF EFFICIENT REGRESSION MODELS FOR APPLIED AGRICULTURAL ECONOMICS RESEARCH
4. The name is absent
5. Word searches: on the use of verbal and non-verbal resources during classroom talk
6. AGRICULTURAL TRADE LIBERALIZATION UNDER NAFTA: REPORTING ON THE REPORT CARD
7. A Bayesian approach to analyze regional elasticities
8. Labour Market Institutions and the Personal Distribution of Income in the OECD
9. Review of “The Hesitant Hand: Taming Self-Interest in the History of Economic Ideas”
10. The name is absent