08.01.2014 Views

Final Report on RREF 2001 - Department of Health

Final Report on RREF 2001 - Department of Health

Final Report on RREF 2001 - Department of Health

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.

Part B: Development <strong>of</strong> Opti<strong>on</strong>s for a Revised <strong>RREF</strong><br />

4.6 FURTHER ISSUES AND PREFERRED OPTIONS<br />

Two aspects <strong>of</strong> differential weighting <strong>of</strong> the need variables were canvassed<br />

in the c<strong>on</strong>sultati<strong>on</strong>s:<br />

1. views as to the variables that warranted greater weight <strong>on</strong> the basis <strong>of</strong><br />

making a greater c<strong>on</strong>tributi<strong>on</strong> to need for HACC services, and<br />

2. the actual weight that each variable c<strong>on</strong>tributed to the model.<br />

Variables warranting greater weighting<br />

Participants in the Round 2 c<strong>on</strong>sultati<strong>on</strong>s expressed the view that some variables were seen to<br />

have a greater bearing <strong>on</strong> need than others. The two variables that were most widely seen to<br />

warrant more weight were rurality and socio-ec<strong>on</strong>omic status, each identified in around 40% <strong>of</strong><br />

resp<strong>on</strong>ses, followed by the health indicator and the residential care shortfall variable, each<br />

identified in around 20% <strong>of</strong> resp<strong>on</strong>ses. Only a minority <strong>of</strong> resp<strong>on</strong>ses, around 10%, identified<br />

cultural diversity and the Koori variable as warranting additi<strong>on</strong>al weighting.<br />

At the same time, it was recognised that there were no clear empirical grounds for adjusting the<br />

weightings and it was noted that any weighting based <strong>on</strong> arbitrary decisi<strong>on</strong>s would complicate<br />

the process; even those who favoured giving additi<strong>on</strong>al weighting to some variables commented<br />

that the process was imprecise. The opti<strong>on</strong> <strong>of</strong> further statistical modelling bey<strong>on</strong>d the principal<br />

comp<strong>on</strong>ents analysis already carried out was precluded by the lack <strong>of</strong> a measure <strong>of</strong> need that<br />

could be used as the independent variable in methods such as regressi<strong>on</strong> analysis.<br />

Actual c<strong>on</strong>tributi<strong>on</strong> <strong>of</strong> variables to model<br />

The scaling <strong>of</strong> all variables from 0 to 9 treats each variable as having an equal relati<strong>on</strong>ship to<br />

need for HACC services. This equal treatment or equal weighting does not however mean that<br />

each variable does in fact make the same c<strong>on</strong>tributi<strong>on</strong> to overall need. The c<strong>on</strong>tributi<strong>on</strong> that<br />

each variable makes depends <strong>on</strong> the distributi<strong>on</strong> <strong>of</strong> LGAs across the range <strong>of</strong> scores from 0 to 9.<br />

A variable <strong>on</strong> which many LGAs receive a high scaled score will make a greater c<strong>on</strong>tributi<strong>on</strong> to<br />

the overall outcome than <strong>on</strong>e <strong>on</strong> which <strong>on</strong>ly a few LGAs have a high scaled score and most<br />

have lower scores.<br />

Thus, as a large number <strong>of</strong> LGAs have a relatively high scaled score <strong>on</strong> median household<br />

income, reflecting low socio-ec<strong>on</strong>omic status, this variable c<strong>on</strong>tributes a large part <strong>of</strong> the overall<br />

model. Similarly, a large number <strong>of</strong> LGAs receive a relatively high YLD and ARIA scores. In<br />

c<strong>on</strong>trast, <strong>on</strong>ly a small number <strong>of</strong> LGAs have a high proporti<strong>on</strong> <strong>of</strong> their populati<strong>on</strong> that is Koori<br />

and many LGAs receive a low scaled score <strong>on</strong> this variable.<br />

The c<strong>on</strong>tributi<strong>on</strong> that each variable makes to the revised <strong>RREF</strong> model is shown in Table 4.1 in<br />

which the total points that each variable actually c<strong>on</strong>tributes are also expressed as a percentage<br />

<strong>of</strong> the actual total <strong>of</strong> 1293.<br />

Even without any further adjustment to take account <strong>of</strong> perceived relati<strong>on</strong>ships to need, it is<br />

evident that each variable makes a different c<strong>on</strong>tributi<strong>on</strong> to the overall model and so carries a<br />

different weight. Thus, the Household Income variable c<strong>on</strong>tributes 494 points, 38.2% <strong>of</strong> the<br />

total actual points, while the cultural diversity variable c<strong>on</strong>tributes 136 points and 10.5% <strong>of</strong> total<br />

need. Given these differences, it may be more appropriate to refer to “unadjusted” weightings<br />

rather than “even” weightings.<br />

<str<strong>on</strong>g>Final</str<strong>on</strong>g> <str<strong>on</strong>g>Report</str<strong>on</strong>g> July <strong>2001</strong> 46

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

Saved successfully!

Ooh no, something went wrong!