Introduction

In this article I aim to provide some visualisations and analysis of the star rating extract for Q3 FY22-23, released by the Australian Government Department of Health and Aged Care (2023). I show to what extent ratings differ by factors such as organisation purpose, organisation size, state, and remoteness. I also present how star ratings have changed since Q2 FY22-23.

Overview

The number of services given each star rating in each category is summarised, for Q3 FY22-23 in Table 1 and Q2 FY22-23 in Table 2.

Table 1: Number of services that were given each star rating within each category in Q3 FY22-23.

Rating	Overall	Residents’ experience	Compliance	Staffing	Quality measures
★★★★★	56	65	1406	225	574
★★★★	1110	885	480	225	647
★★★	1230	1384	552	629	1025
★★	62	178	33	945	163
★	6	2	6	462	77

Table 2: Number of services that were given each star rating within each category in Q2 FY22-23.

Rating	Overall	Residents’ experience	Compliance	Staffing	Quality measures
★★★★★	54	50	1373	240	579
★★★★	967	737	476	211	612
★★★	1358	1494	610	578	1017
★★	115	234	59	934	161
★	6	5	6	574	163

The same information is also visualised as a bar chart in Figure 1.

Most changes are fairly minor, but overall star ratings have tended to increase with a noticeable decrease in two star ratings.
Residents’ experience results appear to be a driving factor in the increase in star ratings as the survey has now been conducted again for many services.
Compliance ratings haven’t changed significantly, which isn’t very notable since these are generally modified on a rolling basis as audits are conducted.
There is a reduction in one star ratings for staffing.
The overall distribution of quality measures ratings does not appear to have changed much, but there is a notable reduction in one star ratings. This is likely to be due to fewer homes failing to submit data, which would otherwise automatically result in a one star rating.

Figure 1: Number of services that were given each star rating within each category in Q2 FY22-23 and Q3 FY22-23.

Table of star ratings

The following table contains the average star rating in each category grouped by providers. Each service’s individual star ratings can be viewed by expanding the provider rows. The unrounded column corresponds to the unrounded overall star rating as described in this article.

Trends

Visually it appears that star ratings have generally been increasing, but to formalise the observation I consider the average star rating. Taking the average of star ratings makes sense if distances between adjacent star ratings are equivalent, e.g., from 1 to 2 should be equivalent to from 4 to 5. Valid arguments can be made for and against this assumption, but I consider it reasonable enough. The average (mean) for each quarter is visualised in Figure 2. Note that the y-axis is not shared between plots, but all categories of star ratings had increasing averages.

Figure 2: Average of the star ratings, along with a 95% confidence interval, for each star rating category in Q2 FY22-23 and Q3 FY22-23.

To determine whether these changes are also statistically meaningful I performed two-sample t-tests for each star rating category. The results are summarised in Table 3. Within each star rating category there is evidence at a 5% level that the average star rating increased from the previous quarter.

Table 3: The results of performing a two sample t-test for each category of star ratings between Q2 FY22-23 and Q3 FY22-23.

Category	Mean Q2	Mean Q3	Difference 95% CI	p-value
Overall	3.379	3.466	[0.053, 0.121]	<0.001
Residents’ experience	3.235	3.331	[0.06, 0.132]	<0.001
Compliance	4.248	4.311	[0.013, 0.112]	0.014
Staffing	2.452	2.520	[0.003, 0.133]	0.041
Quality measures	3.507	3.595	[0.029, 0.146]	0.003

Pattern of change

Instead of just considering the overall change it may be interesting to look at how many services moved from having a low rating to a high rating and vice versa. This can indicate how high mobility is in a particular star rating category, or inversely, how strong the correlation is between subsequent quarters. If correlation is very high then that might indicate the star ratings are rarely updated or homes’ ratings don’t change much. Conversely, a low correlation may indicate that the rating is less sticky and is updated more frequently. However, a very low correlation could suggest a rating is not reliable since it is unlikely that a large proportion of aged care services changed significantly on a quarterly basis.

The count of star ratings that changed by each amount is visualised in Figure 3.

Figure 3: Change in star rating from Q2 FY22-23 to Q3 FY22-23. Positive values represents an increase.

The changes can also be visualised as a heat map, as in Figure 4, where colour represent the frequency. This chart is quite informative for exploring the patterns of star rating changes. 1. There were no homes that went from having a one star overall rating to a four or five star overall rating. 2. The diagonal is darker, representing the tendency of star ratings to stay the same. 3. The tiles above the diagonal are generally darker than below the diagonal signalling an upward trend in star ratings.

Figure 4: Heatmap of homes’ star ratings in Q2 FY22-23 and Q3 FY22-23.

The proportion of star ratings that changed from the previous quarter, for each category, is shown in Figure 5.

Figure 5: Proportion of star ratings that changed within each star rating category.

Sankey charts are a nice way to visualise the most common pathways from one quarter to another. The Sankeys for overall, residents’ experience, compliance, and quality measure star ratings are shown in Figure 6, Figure 7, Figure 8, Figure 9, and Figure 10, respectively. It is similar to the heatmap in what information can be ascertained. The star ratings for Q2 FY22-23 are represented by rectangles on the left and star ratings for Q3 FY22-23 are represented by rectangles on the right. The flows show the number of homes that went from one particular star rating to another. In a desktop browser the flows of the Sankey can be hovered over to get a count, if this doesn’t work refer to the heatmap instead.

Figure 6: Overall star rating changes from Q2 FY22-23 to Q3 FY22-23.

Figure 7: Residents’ experience star rating changes from Q2 FY22-23 to Q3 FY22-23.

Figure 8: Compliance star rating changes from Q2 FY22-23 to Q3 FY22-23.

Figure 9: Staffing star rating changes from Q2 FY22-23 to Q3 FY22-23.

Figure 10: Quality measures star rating changes from Q2 FY22-23 to Q3 FY22-23.

Explanatory factors

In this section I consider the association between various factors provided in the star rating extracts that may be associated with higher or lower ratings.

Summary

The mean of each star rating category stratified by various factors is summarised in Table 4.

Table 4: The mean of each star rating category stratified by various factors.

	Group	Overall	RE	Compliance	Staffing	QM	N
Organisation purpose	Not for profit	3.43	3.38	4.25	2.41	3.62	1505
Organisation purpose	For profit	3.38	3.19	4.35	2.28	3.54	905
Organisation purpose	Government	4.03	3.59	4.54	4.17	3.67	214
Organisation size	Small	3.53	3.40	4.20	2.71	3.72	940
Organisation size	Medium	3.46	3.29	4.33	2.61	3.53	775
Organisation size	Large	3.40	3.29	4.42	2.23	3.52	909
State	NSW	3.41	3.37	4.14	2.43	3.63	830
State	VIC	3.53	3.24	4.44	2.69	3.66	745
State	QLD	3.55	3.48	4.53	2.46	3.40	465
State	WA	3.33	3.24	4.23	2.38	3.42	248
State	SA	3.42	3.34	4.12	2.54	3.68	230
State	ACT	3.41	3.15	4.22	2.52	3.96	27
State	TAS	3.44	3.26	4.42	2.40	3.91	70
State	NT	3.62	2.88	4.00	3.56	4.00	9
Modified Monash Model	MM1	3.44	3.23	4.34	2.42	3.61	1649
Modified Monash Model	MM2	3.44	3.38	4.36	2.40	3.48	209
Modified Monash Model	MM3	3.33	3.41	4.10	2.33	3.57	229
Modified Monash Model	MM4	3.44	3.45	4.24	2.66	3.47	189
Modified Monash Model	MM5	3.73	3.67	4.34	3.05	3.72	311
Modified Monash Model	MM6	3.67	3.33	4.36	3.32	3.54	28
Modified Monash Model	MM7	3.67	3.67	4.11	3.44	3.11	9
Modified Monash Model grouped	Metropolitan (MM1)	3.44	3.23	4.34	2.42	3.61	1649
Modified Monash Model grouped	Regional centres (MM2)	3.44	3.38	4.36	2.40	3.48	209
Modified Monash Model grouped	Rural and remote (MM3-7)	3.54	3.53	4.24	2.76	3.60	766

Organisation purpose

The number of services of each organisation type is shown in Table 5. Figure 11 shows the proportion of services that received each star rating and the average star rating along with a 95% confidence interval, stratified by organisation type and star rating category.

The 95% estimated confidence interval, within each star rating category, was calculated as $\overline{x} \pm 1.96 s / \sqrt{n}$ , where $\overline x$ is the sample mean, $s$ is the sample standard deviation, and $n$ is the sample size.

Table 5: The number of aged care services of each organisation purpose with non-missing overall star rating.

Organisation type	Count
For profit	849
Government	210
Not for profit	1405

Figure 11: The proportion of services that received each star rating within each category stratified by organisation purpose and the mean star rating along with an estimated 95% confidence interval.

Organisation size

I categorised provider size by the number of unique service names for the provider name in the extract: small (1-4), medium (5-19), large (20+). This may not perfectly capture the size of organisations as there are many business structures in the aged care industry, so the variable should be considered with this limitation in mind. The number of services that are part of an organisation of each size is shown in Table 6. Figure 12 shows the proportion of services that received each star rating and the average star rating along with a 95% confidence interval, stratified by organisation size and star rating category.

Table 6: The number of aged care services of each organisation size with non-missing overall star rating.

Organisation size	Count
Small	903
Medium	736
Large	825

Figure 12: The proportion of services that received each star rating within each category stratified by organisation size and the mean star rating along with an estimated 95% confidence interval.

Geography

Remoteness is described by the Modified Monash Model (MMM), which divides Australia into seven categories of remoteness from MM1 to MM7. I use MMM in its original form along with a grouping of MM1, MM2, and MM3-7. I also consider the Australian states and territories. The number of services in each group and how these relate to each other are shown in the alluvial chart in Figure 13.

The geographical categories introduced above are visualised in an interactive map in Figure 14.

Figure 14: An interactive map showing a simplified version of the geographical boundaries of interest.

Figure 15, Figure 16, and Figure 17 show the proportion of services that received each star rating and the average star rating along with a 95% confidence interval, stratified by geographical groups. It appears that staffing ratings are higher in more remote locations. Residents’ experience ratings are also increasing from MM1 to MM5.

Figure 15: The proportion of services that received each star rating within each category stratified by States and territories and the mean star rating along with an estimated 95% confidence interval.

Figure 16: The proportion of services that received each star rating within each category stratified by the Modified Monash Model and the mean star rating along with an estimated 95% confidence interval.

Figure 17: The proportion of services that received each star rating within each category stratified by a grouped Modified Monash Model and the mean star rating along with an estimated 95% confidence interval.

Stratified trend

The average star ratings along with 95% confidence intervals stratified by the explanatory factors, organisation purpose, organisation size, and remoteness, and some states are visualised in Figure 18, Figure 19, Figure 20, and Figure 21. Corresponding t-tests for differences between Q2 FY22-23 and Q3 FY22-23 star ratings stratified by each factor are shown in Table 7, Table 8, Table 9, and Table 10.

Figure 18: Average of the star ratings, along with a 95% confidence interval, for each organisation purpose in Q2 FY22-23 and Q3 FY22-23.

Table 7: The results of performing a two sample t-test for each category of star rating and organisation purpose between Q2 FY22-23 and Q3 FY22-23.

Category	Group	Mean Q2	Mean Q3	Difference 95% CI	p-value
Overall	For profit	3.280	3.383	[0.049, 0.157]	<0.001
	Government	3.963	4.033	[-0.047, 0.188]	0.238
	Not for profit	3.350	3.431	[0.038, 0.124]	<0.001
Residents' experience	For profit	3.086	3.193	[0.049, 0.165]	<0.001
	Government	3.581	3.590	[-0.124, 0.143]	0.885
	Not for profit	3.272	3.377	[0.059, 0.152]	<0.001
Compliance	For profit	4.308	4.352	[-0.037, 0.126]	0.287
	Government	4.458	4.537	[-0.077, 0.235]	0.320
	Not for profit	4.182	4.251	[0.002, 0.136]	0.045
Staffing	For profit	2.265	2.281	[-0.075, 0.108]	0.721
	Government	4.092	4.173	[-0.157, 0.32]	0.504
	Not for profit	2.317	2.414	[0.018, 0.177]	0.017
Quality measures	For profit	3.385	3.541	[0.058, 0.253]	0.002
	Government	3.771	3.668	[-0.305, 0.1]	0.320
	Not for profit	3.539	3.616	[-0.002, 0.155]	0.056

Figure 19: Average of the star ratings, along with a 95% confidence interval, for each organisation size in Q2 FY22-23 and Q3 FY22-23.

Table 8: The results of performing a two sample t-test for each category of star rating and provider size between Q2 FY22-23 and Q3 FY22-23.

Category	Group	Mean Q2	Mean Q3	Difference 95% CI	p-value
Overall	Small	3.400	3.530	[0.071, 0.189]	<0.001
	Medium	3.394	3.458	[0, 0.128]	0.051
	Large	3.341	3.402	[0.008, 0.114]	0.023
Residents' experience	Small	3.288	3.402	[0.052, 0.174]	<0.001
	Medium	3.233	3.291	[-0.008, 0.125]	0.085
	Large	3.176	3.291	[0.058, 0.174]	<0.001
Compliance	Small	4.138	4.196	[-0.029, 0.145]	0.192
	Medium	4.259	4.330	[-0.017, 0.159]	0.115
	Large	4.366	4.420	[-0.027, 0.135]	0.194
Staffing	Small	2.635	2.713	[-0.036, 0.193]	0.181
	Medium	2.488	2.610	[-0.001, 0.247]	0.053
	Large	2.204	2.227	[-0.07, 0.116]	0.629
Quality measures	Small	3.589	3.718	[0.027, 0.229]	0.013
	Medium	3.461	3.531	[-0.039, 0.178]	0.209
	Large	3.453	3.517	[-0.03, 0.158]	0.182

Figure 20: Average of the star ratings, along with a 95% confidence interval, for each MMM grouping in Q2 FY22-23 and Q3 FY22-23.

Table 9: The results of performing a two sample t-test for each category of star rating and MMM grouping between Q2 FY22-23 and Q3 FY22-23.

Category	Group	Mean Q2	Mean Q3	Difference 95% CI	p-value
Overall	Metropolitan (MM1)	3.334	3.436	[0.062, 0.143]	<0.001
	Regional centres (MM2)	3.396	3.437	[-0.073, 0.154]	0.484
	Rural and remote (MM3-7)	3.470	3.536	[-0.003, 0.137]	0.062
Residents' experience	Metropolitan (MM1)	3.122	3.235	[0.07, 0.156]	<0.001
	Regional centres (MM2)	3.267	3.377	[-0.006, 0.225]	0.063
	Rural and remote (MM3-7)	3.465	3.527	[-0.007, 0.131]	0.078
Compliance	Metropolitan (MM1)	4.288	4.337	[-0.012, 0.11]	0.112
	Regional centres (MM2)	4.255	4.355	[-0.072, 0.273]	0.252
	Rural and remote (MM3-7)	4.165	4.244	[-0.018, 0.176]	0.112
Staffing	Metropolitan (MM1)	2.353	2.421	[-0.005, 0.142]	0.068
	Regional centres (MM2)	2.426	2.402	[-0.278, 0.229]	0.850
	Rural and remote (MM3-7)	2.667	2.758	[-0.046, 0.229]	0.194
Quality measures	Metropolitan (MM1)	3.487	3.608	[0.049, 0.194]	0.001
	Regional centres (MM2)	3.490	3.477	[-0.227, 0.201]	0.906
	Rural and remote (MM3-7)	3.553	3.597	[-0.068, 0.156]	0.442

Figure 21: Average of the star ratings, along with a 95% confidence interval, for select states in Q2 FY22-23 and Q3 FY22-23.

Table 10: The results of performing a two sample t-test for each category of star rating and state between Q2 FY22-23 and Q3 FY22-23.

Category	Group	Mean Q2	Mean Q3	Difference 95% CI	p-value
Overall	ACT	3.185	3.407	[-0.104, 0.549]	0.178
	NSW	3.319	3.410	[0.03, 0.151]	0.003
	NT	3.000	3.625	[-0.078, 1.328]	0.077
	QLD	3.436	3.554	[0.043, 0.192]	0.002
	SA	3.317	3.423	[-0.012, 0.224]	0.079
	TAS	3.429	3.439	[-0.19, 0.212]	0.915
	VIC	3.491	3.532	[-0.025, 0.106]	0.221
	WA	3.214	3.330	[0.011, 0.222]	0.031
Residents' experience	ACT	2.889	3.148	[-0.026, 0.544]	0.073
	NSW	3.284	3.368	[0.018, 0.15]	0.013
	NT	2.750	2.875	[-0.599, 0.849]	0.717
	QLD	3.238	3.484	[0.167, 0.324]	<0.001
	SA	3.258	3.344	[-0.027, 0.199]	0.136
	TAS	3.229	3.258	[-0.176, 0.234]	0.780
	VIC	3.201	3.238	[-0.03, 0.104]	0.281
	WA	3.210	3.241	[-0.088, 0.151]	0.608
Compliance	ACT	4.148	4.222	[-0.44, 0.588]	0.774
	NSW	4.080	4.138	[-0.036, 0.151]	0.229
	NT	3.917	4.000	[-0.737, 0.904]	0.834
	QLD	4.431	4.532	[-0.001, 0.204]	0.052
	SA	4.019	4.120	[-0.079, 0.281]	0.270
	TAS	4.357	4.424	[-0.226, 0.36]	0.651
	VIC	4.405	4.442	[-0.049, 0.124]	0.394
	WA	4.196	4.235	[-0.128, 0.205]	0.649
Staffing	ACT	2.333	2.519	[-0.406, 0.776]	0.532
	NSW	2.304	2.433	[0.028, 0.231]	0.013
	NT	2.667	3.556	[-0.46, 2.238]	0.184
	QLD	2.444	2.459	[-0.135, 0.164]	0.848
	SA	2.426	2.541	[-0.096, 0.326]	0.285
	TAS	2.457	2.403	[-0.45, 0.341]	0.787
	VIC	2.678	2.691	[-0.132, 0.157]	0.865
	WA	2.311	2.385	[-0.108, 0.255]	0.426
Quality measures	ACT	3.741	3.963	[-0.292, 0.737]	0.390
	NSW	3.553	3.627	[-0.027, 0.176]	0.148
	NT	3.167	4.000	[-0.474, 2.141]	0.198
	QLD	3.452	3.401	[-0.18, 0.078]	0.437
	SA	3.550	3.679	[-0.055, 0.314]	0.170
	TAS	3.786	3.910	[-0.282, 0.531]	0.545
	VIC	3.633	3.662	[-0.081, 0.139]	0.605
	WA	2.963	3.423	[0.243, 0.676]	<0.001

Unrounded overall star rating

I calculated the unrounded overall star ratings according to the Star Ratings Provider Manual (Australian Government, Department of Health and Aged Care 2022). The overall star rating is calculated according to the formula, $\begin{align*} \text{Overall star rating} &= 0.33\times \text{Resident's experience star rating}\\ &\quad +0.30\times \text{Compliance star rating} \\ &\quad +0.22\times \text{Staffing star rating} \\ &\quad +0.15\times \text{Quality measures star rating}. \end{align*}$ Except when the compliance star rating is two or lower, then the overall star rating can be no greater than the compliance rating. The final step would be to round to the nearest whole number.

Analysing the star ratings prior to the final step of rounding the overall star rating to a whole number can be insightful. In some sense this quantity contains more information than the final overall star ratings. Additionally, unrounded star ratings have the property of resembling a familiar statistical distribution, the normal distribution. This is clear from Figure 22. The graph also shows that the upward shift in star ratings is fairly consistent across the distribution of star ratings. The masses at 1 and 2 star ratings is due to the “compliance rule” step of the algorithm mentioned earlier.

Figure 22: The distribution of unrounded star ratings in each quarter.

The distribution of differences between unrounded star ratings in matching records for Q2 FY22-23 and Q3 FY22-23 are visualised in Figure 23.

Figure 23: The distribution of changes in unrounded star ratings from Q2 FY22-23 to Q3 FY22-23. Positive values represents an increase.

There are many small changes in star ratings that did not result in a change in the overall star rating. This phenomenon is visualised in Figure 24. Since the overall star rating is rounded, there is a range of unrounded overall star ratings that will result in the same star rating.

Figure 24: The unrounded overall star ratings for Q2 FY22-23 compared to for Q3 FY22-23. The colour indicates whether the rounded overall rating changed.

If instead the unrounded changes are of more interest, the magnitude of changes can be visualised as in Figure 25.

Figure 25: The unrounded overall star ratings for Q2 FY22-23 compared to for Q3 FY22-23. The colour indicates the magnitude and direction of change in the unrounded overall star rating.

The unrounded star ratings are also useful for considering the effect of explanatory factors. I performed an analysis of which factors matter for star ratings in the post Multivariable Linear Model for Aged Care Star Ratings. Instead of repeating the analysis, I provide some summary statistics of the unrounded overall star rating stratified by factors considered earlier in this article. Table 11 shows the five point summary, mean, standard deviation, sample size, density plot, and box plot for each stratum.

Table 11: Summary statistics, a density plot, and a boxplot for the unrounded overall star rating stratified by various factors.

	Min	Q1	Median	Q3	Max	Mean	SD	N
Organisation type
Not for profit	1.0	3.2	3.4	3.8	5.0	3.4	0.5	1405
For profit	1.0	3.1	3.4	3.7	4.7	3.4	0.5	848
Government	2.0	3.7	4.1	4.4	4.8	4.0	0.5	210
Organisation size
Small	1.0	3.2	3.5	3.9	5.0	3.5	0.5	903
Medium	1.0	3.2	3.4	3.8	4.8	3.5	0.5	735
Large	1.0	3.2	3.4	3.7	4.7	3.4	0.4	825
State
NSW	1.0	3.1	3.4	3.7	4.7	3.4	0.5	787
VIC	2.0	3.2	3.5	3.9	4.8	3.5	0.5	697
QLD	2.4	3.3	3.5	3.8	5.0	3.6	0.4	446
WA	1.0	3.1	3.4	3.7	4.5	3.4	0.5	224
SA	1.0	3.1	3.4	3.8	4.7	3.4	0.5	208
ACT	2.0	3.2	3.5	3.7	4.4	3.4	0.5	27
TAS	2.0	3.3	3.4	3.8	4.7	3.5	0.4	66
NT	2.8	3.2	3.6	3.8	4.1	3.5	0.4	8
Modified Monash Model
MM1	1.0	3.2	3.4	3.7	5.0	3.4	0.5	1538
MM2	2.4	3.2	3.4	3.8	4.7	3.5	0.4	197
MM3	1.0	3.1	3.4	3.7	4.7	3.4	0.5	212
MM4	1.0	3.2	3.5	3.9	4.8	3.5	0.6	183
MM5	1.0	3.4	3.8	4.1	4.8	3.7	0.6	297
MM6	2.8	3.4	3.8	3.9	4.4	3.7	0.5	27
MM7	3.0	3.2	3.5	4.2	4.7	3.7	0.6	9
Modified Monash Model grouped
Metropolitan (MM1)	1.0	3.2	3.4	3.7	5.0	3.4	0.5	1538
Regional centres (MM2)	2.4	3.2	3.4	3.8	4.7	3.5	0.4	197
Rural and remote (MM3-7)	1.0	3.2	3.5	3.9	4.8	3.6	0.6	728

Notes on data

Columns

There are new columns available that describe the star ratings in more detail, however I didn’t analyse these here. The home size column was available in Q2 FY22-23, but is not available for Q3 FY22-23, so it was not available for this analysis.

Missing data

Some star ratings are missing in the dataset. Based on the Star Ratings Provider Manual (Australian Government, Department of Health and Aged Care 2022) there are a number of reasons that a star rating may not be available such as: the service has not operated for two reporting quarters of the National Aged Care Mandatory Quality Indicator Program; the service has recently gone through a transfer in ownership; the rating for quality measures or staffing is under review by the department; unavailability of data. Furthermore, the overall star rating can generally only be determined when all other star ratings are available for that service. The number of star ratings missing in each category is shown in Table 12. The patterns of missingness are shown in Table 13. We can confirm from this table that the overall star rating is generally only available when all other star ratings are also available. However, there is one exception, which is the result of a one star compliance rating causing the overall rating to be 1, no matter the residents’ experience rating.

Table 12: The number of star ratings unavailable within each category.

	Unavailable star ratings
Overall	160
Residents’ experience	110
Compliance	147
Staffing	138
Quality measures	138

Table 13: Number of services with each combination of missing star ratings.

Missing star rating for	Count
Overall, Residents’ experience, Compliance, Staffing, Quality measures	94
Overall, Compliance, Staffing, Quality measures	40
Overall, Residents’ experience	9
Overall, Compliance	8
Overall, Residents’ experience, Compliance	4
Overall, Staffing, Quality measures	2
Overall, Quality measures	1
Overall, Residents’ experience, Compliance, Staffing	1
Overall, Residents’ experience, Staffing, Quality measures	1
Residents’ experience	1

Data linkage

Unfortunately, the overlap between the Q2 FY22-23 extract and Q3 FY22-23 extract is not perfect. Aged care homes can be opened, change name, transfer ownership, or be closed down. Some homes also have two records due to having a low and high care wing. If I only keep records that are directly matched in the extracts then the overall average star rating is 3.364 (instead of 3.379) for Q2 FY22-23 and 3.444 (instead of 3.466) for Q3 FY22-23. Based on this and similar sensitivity analysis among other categories (not shown), I decided to use records where the non-star rating columns matched between the data extracts, whenever linking was required. Among records in the extract for Q3 FY22-23, 97.1% had a matching record for Q2 FY22-23. However, whenever linking between the datasets was not required I used the full unmodified data extract for each quarter.

References

Australian Government, Department of Health and Aged Care. 2022. “Star Ratings Provider Manual.” https://www.health.gov.au/resources/publications/star-ratings-provider-manual?language=en.

Australian Government Department of Health and Aged Care. 2023. “Star Ratings Quarterly Data Extracts.” https://www.health.gov.au/resources/collections/star-ratings-quarterly-data-extracts.