The past decade has seen an international intensification of quality rating and improvement systems (QRIS) for early childhood education and care (ECEC) services as a key feature of ensuring that all children “have access to quality early childhood development, care and pre-primary education” (United Nations Sustainable Development Goal 4, Target 4.2, 2015). While the primary role of National, State and regional regulatory systems in underpinning the structural determinants of high-quality educational practices (Cryer et al., 1999; Phillipsen et al., 1997) is well understood and accepted, the nuances of the layered structural arrangements of ECEC services in relation to process quality are yet to be fully explored. In this paper, we extend “systems thinking” (Kagan & Roth, 2017, p. 146) to examine the impact of organisational structures and systemic features on QRIS outcomes in a large sample of Australian ECEC centres. We propose the term ‘macro-structural’ to capture the multi-layered structural arrangements that influence quality improvement.

Slot (2018) identified four distinct levels of structural characteristics: systemic, organisational, classroom and staff, as predictors of ECEC process quality. She proposed the need for thinking that moves beyond “the iron triangle characteristics” of child: staff ratio (numbers of children per adult), group size (number of children in a classroom) and staff qualifications, and for research that aims to identify and include structural features at both the systemic/policy level and the organisational/service level (p. 8). Slot further suggested taking account of and testing complex interactions between and within different system levels.

This paper examines the combined influences of seven macro-structural features quality improvement in centre-based ECEC. At the systems level, we distinguish (1) state/territory jurisdictional responsibilities for ECEC quality and contextual aspects of (2) location/degree of urbanisation and (3) community socio-economic status (SES). At the organisational level, we include (4) governance structure (e.g. not-for-profit, for-profit) and (5) size of the organisation that owns or operates the centre. Organisational features also include (6) centre size (i.e. number of child places) and (7) operational stability of the centre (i.e. change of ownership). Finally, we examine previous research, noting studies that focus on one or a combination of these seven features.

Systemic influences on quality in ECEC

According to the Organisation for Economic Cooperation and Development (OECD) (2019), ECEC policy in various countries is governed by integrated centralised systems or by spilt systems. Globally, the integration of ECEC services within one jurisdictional level (that is, when a singular government ministry or authority is responsible for the operation of ECEC services in terms of regulations, funding and conceptual ideas) has been linked to higher quality in ECEC services (OECD, 2019). Where split systems exist, the division tends to be around the education and care of the youngest children (birth up to 3 years) versus preschool-age children (3–5 years) (OECD, 2019). About 50% of OECD countries have a unitary system for the age range from birth to 5, or sometimes until 8 years of age (Guerriero, 2017). In these countries, the Ministry of Education is the responsible governing body. In countries with a split system, education authorities are usually responsible for preschool provisions for children aged 3 years and up, whereas social services authorities are responsible for child care provisions for children aged birth to 2 years (Guerriero, 2017; Slot, 2018).

Countries with split systems may also have decentralised ECEC policies that operate at the state or provincial level. For example, in the United States, there are wide variations in the minimum requirements for teacher qualifications (from high school diploma to preservice teaching credentials) and staff-to-child ratios (from 1:7 to 1:15 for 3-year-old children) that are set by the individual states (NCECQA, 2015). State-based licencing provisions also reflect differences in mechanisms for monitoring quality, the stringency of these regulatory mechanisms, financing and subsidy flow, and professional development opportunities for educators that can influence quality outcomes (Connors & Morris, 2015; Maxwell & Starr, 2019). Australia, in 2012, changed from a split to an integrated system, by adopted a jointly governed, uniform and integrated national approach to ECEC quality monitoring and improvement for children from birth to 5 years (Australian Children’s Education and Care Quality Authority [ACECQA], 2020). Prior to the establishment of the National Quality Framework (NQF), Australian states and territories had different licensing requirements with different minimum standards for staff qualifications and staff–child ratios (Dowling & O’Malley, 2009). The implementation of the NQF included transitional arrangements for the states/territories to achieve the agreed new requirements for educator qualifications and child ratios.

Location—degree of urbanisation

Research conducted in the United States and China has noted distinctions in quality indicators between ECEC services located in metropolitan vs rural areas. Maher et al.'s (2008) examination of urban–rural differences in five US states demonstrated that rural areas were more likely to have higher child-to-adult ratios for infants and lower ratios for preschoolers than urban centres. In a study of program quality in Eastern China using the Chinese Early Childhood Programs Rating Scale, Hu et al. (2014) identified a significant urban–rural gap that was affected by the type of governance structure: urban public (not-for-profit) kindergartens received the highest quality ratings, followed by urban private (for-profit) and rural public, with rural private kindergartens receiving the lowest ratings. A further study of quality in Western China found that ECEC in rural areas was under-funded, rural teachers were under-compensated, and public kindergartens had high child-to-adult ratios (Zhou et al., 2017). In contrast, Thorpe et al.’s (2021) analysis of a national sample of Australian ECEC services reported very little variation in quality ratings by geographic area.

Community SES

The socio-economic status of the community where an ECEC service is situated can also make a difference. For example, Hatfield et al. (2015) found that the SES of communities in North Carolina was related to program quality as measured by the Tiered QRIS. Similarly, Bassok & Galdo (2016) identified disparities in access to and the quality of early learning opportunities in public preschools in Georgia: program quality was lower in low-income and high-minority communities than in higher-income communities. In Australia, Cloney et al. (2016) examined the impact of neighbourhood SES on quality in 421 preschool classrooms in Victoria and Queensland, using the prekindergarten version of the Classroom Assessment Scoring System (CLASS). Results confirmed that the lowest-SES communities received lower ratings on all CLASS domains compared to classrooms in the highest-SES communities. However, a national comparison, based on the Australian QRIS, reported “relative parity” across socio-economic indices for communities (Thorpe et al., 2021, p. 233).

Governance structure of provider organisations

Organisations that provide ECEC services are broadly grouped as for-profit or not-for-profit. According to Ünver et al. (2018), for-profit services provide market-based heterogeneity through encouraging competition and are “expected to result in cheaper, more efficient, responsive, innovative, flexible and high-quality services” (p. 4). International evidence, however, suggests quality is lower in for-profit services. In Australia, quarterly reports produced by ACECQA (2022) consistently show lower levels of quality for services operated by for-profit providers (see also Thorpe et al., 2021). In New Zealand, Mitchell (2012) has suggested that some organisations that provide for-profit ECEC services perceive staff wages as an infringement on profits, with attempts to reduce these costs by employing staff with minimal qualifications and resulting in conditions that are not conducive to staff motivation and teaching. Similar concerns have been raised by Penn (2011) in her analysis of for-profit services in the United Kingdom. In contrast, not-for-profit organisations in Canada, the United States, United Kingdom and Europe have been found to achieve greater staff motivation and capacity to provide high-quality programs by ensuring higher educator–child ratios, higher staff qualifications and remuneration, and better professional development opportunities (Aubrey, 2019; Cleveland & Krashinsky, 2009; Coley et al., 2016; Slot, 2018).

Provider organisation size

In many countries, the ECEC sector operates as a mixed-market system notable for its diversity, ranging from single-service providers to large organisations that operate multi-site centres. Diversity in provider size is characteristic of both for-profit and not-for-profit services in countries such as Australia (Brennan & Fenech, 2014), Canada (Varmuza et al., 2021) and Finland (Mäntyjärvi & Puroila, 2019), with some research suggesting that provider size can impact quality. For example, a national survey of Australian educators showed that centres that were part of a large corporate chain were rated significantly lower on having time to develop individual relationships with children than centres operated by small private operators (Rush, 2006). In addition, observations of positive caregiving gathered in a large US study have shown lower scores in for-profit multi-site organisations compared to independent for-profit services, for some but not all age groups (Sosinsky et al., 2007). More recently, Varmuza et al. (2021) have examined three years of Canadian QRIS data for single and multi-site, for-profit and not-for-profit providers using the Assessment for Quality Improvement (AQI). While there were no consistent differences in annual comparisons of quality by the size of provider, multi-site for-profit operators showed increasing AQI ratings over time.

Centre size

Differences in size are also a feature of individual ECEC centres, which vary in the number of licensed places for children. While the number of children per adult and per classroom are known indicators of ECEC quality (Cryer et al., 1999), few studies have examined the relationship between the overall number of children and quality. Ho et al. (2016) found that teachers working in smaller services reported higher perceived organisational support, including participation in decision making and support from management, and lower organisational negativity than their counterparts in medium and large settings. In contrast, Wong et al. (2012) noted that centres with fewer enrolled children were negatively impacted by inadequate government financial support.

Centre stability

Staff stability in ECEC services provides certainty for families and trust for children, and is a key determinant of process quality (Irvine et al., 2016; Rush, 2006). Instability, as measured by staff turnover rates, has been associated with lower quality (Sosinsky et al., 2007), but the direction of effect is not clear: Are staff more likely to leave low-quality services, or does staff attrition have a detrimental impact on quality? Thorpe et al. (2021) note that educators working in lower quality services report poorer morale and greater intention to seek better pay and conditions elsewhere than educators working in higher quality services. However, to our knowledge, no prior research has examined the effect of centre stability, as defined by continuity of provider, on quality or quality improvement.

Australia’s quality rating and improvement system

Australia’s National Quality Framework (NQF) established a quality monitoring and improvement system that addresses structural and process quality and spans all service types (long day care, preschool/kindergarten, family day care) (ACECQA, 2020). The NQF includes a National Quality Standard (NQS), and an Assessment and Rating (A&R) process that are underpinned by National Law and National Regulations. Overseen by the national authority ACECQA, each state/territory Regulatory Authority is charged with the day-to-day administration of this national system. The critical distinction between the Australian QRIS and other countries is that participation is mandatory for ECEC services to enable families to access government subsidies for fee support. This policy framing, which is enshrined in legislation, has ensured very high levels of participation.

The A&R process involves the preparation by centres of a Quality Improvement Plan (QIP) that enables services and providers to “self-assess their performance in delivering quality education and care and to plan future improvements” (ACECQA, 2012, p. 34). The QIP is submitted before the A&R review and provides a basis for observations and conversations by the trained external regulatory assessor. The NQS comprises a rigorous process for assessment and rating on seven quality areas (QAs) (see ACECQA, 2020) on a 4-level scale (defined below). There is expert agreement that the underlying constructs of quality in the NQS A&R are aligned with constructs in standardised assessment instruments (Jackson, 2015; Thorpe et al., 2021). At the completion of the A&R process, the service receives a report indicating their QA ratings and an overall rating that is valid for three years, ensuring a sequence of regular re-assessment.

  1. 1)

    Significant Improvement Required: Service does not meet one QA or a section of the legislation, and there is a significant risk to children's safety, health and wellbeing. The regulatory authority will take immediate action;

  2. 2)

    Working Towards NQS: Service provides a safe education and care program. There are one or more QAs identified for improvement;

  3. 3)

    Meeting NQS: Service provides quality education and care in all QAs;

  4. 4)

    Exceeding NQS: Service goes beyond the requirements of the NQS in at least four QAs, with at least two of these being QA1, QA5, QA6 or QA7.

The Australian quality rating and assessment system is designed to promote quality improvement processes in every ECEC service. These expectations go beyond minimum regulations to consider interactive components involving all stakeholders—children, staff, families and the community. Continuous quality improvement is an overarching aim and an integral feature of the NQS A&R review process (ACECQA, 2020).

The current study

The focus of this study was on quality improvement in Australian long day care centres and the macro-structural features that support improved NQS ratings. With few exceptions, prior research has tended to focus on only one or two structural factors, with limited regard for the underlying complexities of the different levels of influence. Further, while QRIS is about quality improvement, prior research has tended to rely on single-point assessments of ECEC quality. We addressed these gaps in the following research question:

Do macro-structural features at three levels—system, organisational and centre—impact long day care quality improvement from one QRIS assessment to the next?

Materials and method

This paper is based on an analysis of QRIS outcomes in long day care centres, made possible by accessing Australia’s national repository of NQS A&R data from its inception in 2012 to December 2018. Secondary analysis of the publicly available NQS A&R administrative data did not involve the collection of any new data by the researchers, and therefore, did not require ethics approval by the University Human Research Ethics Committee.


Data for all long day care centres (N = 3433) that had completed at least two NQS A&R rounds were extracted from the national repository (see Table 1). From this, we extracted a study sample comprising all the centres with an initial rating of Working Towards NQS (N = 1935) that had either improved to Meeting NQS (n = 957; 49.5%), improved to Exceeding NQS (n = 381; 19.7%) or had no change from Working Towards NQS (n = 597; 30.8%) in the follow-up A&R. These assessments occurred between April 2012 and January 2018.

Table 1 Distribution of long day care centres by NQS ratings at time 1 and time 2 assessments


Macro-structural features were extracted from the data repository: three at the systems level (jurisdiction; location—degree of urbanisation and community SES), two at the organisational level (type and size of provider organisation) and two at the centre level (centre size and stability of ownership).


Centres operated under eight state/territory regulatory authorities. New South Wales (NSW), the largest state in Australia, accounted for the largest number of centres (51.8% of the total sample). The distribution across the other seven jurisdictions, in decreasing order, was: Queensland (16.5%), Victoria (12.8%); West Australia (7.0%); South Australia (3.7%); the Australian Capital Territory (3.6%); Northern Territory (2.8%) and Tasmania (1.8%).

Location—degree of urbanisation and community SES

Classification was based on the Australian Bureau of Statistics' (ABS) Accessibility and Remoteness Index of Australia (ARIA +) classes: major cities (metropolitan), inner regional, outer regional, remote and very remote. ARIA + is a general access model covering education, health, shopping, public transport and financial/postal services based on road distance to the nearest service centre locality (Hugo Centre for Population and Migration Studies, 2021). The majority (72.3%) of centres were located in metropolitan areas; 16.9% were located in inner regional areas, 8.6% in outer regional areas and 2.2% in remote/very remote areas.

Community SES was described by the ABS Socio-Economic Index for Areas of Relative Socio-Economic Advantage and Disadvantage (SEIFA), which is based on census data for household income, education, employment, occupation, housing and other indicators. SEIFA ranges from 1 (most disadvantaged) to 10 (most advantaged). For ease of analysis and interpretation, we reduced SEIFA deciles to quintiles: 28.3% of the sample were located in SEIFA 1/2 (more disadvantaged communities), 24.4% in SEIFA 3/4, 18.3% in SEIFA 5/6, 15.4% in SEIFA 7/8 and 13.6% in SEIFA 9/10 (more advantaged communities).

Type and size of provider organisation

The national repository identifies five categories of approved providers: for-profit, not-for-profit community managed, not-for-profit other organisation (non-government), State/Territory and Local Government managed and school managed, including ECEC services within Catholic, Independent and government schools. The majority of centres (70.8%) were operated by for-profit organisations. The rest were operated by not-for-profit community- based organisations (11.2%), non-government organisations (13.8%), State/Territory/Local Government organisations (2.5%) and schools (1.7%). For analysis purposes, we created two categories that described similar governance arrangements: (1) community-based organisations and schools (12.9%); (2) government and non-government organisations (16.3%).

Size of provider organisation was defined by three categories: small—a single stand-alone ECEC service (31.0%); medium—organisations operating 2–7 services (38.8%); and large—organisations operating 8 or more services (30.3%).

Centre size and stability

A cut-off point of 60 licensed places was used to determine small versus large centres, based on the Australian National Regulations that require the number of ECEC-qualified teachers to increase from one to two if a service is licensed for 60 or more children on a given day. The study sample included similar numbers of small (55.1%) and large (44.9%) centres.

Centre stability was based on a reported change of ownership. During the period covered in our study, 35.4% of long day care centres had recorded a transfer from one approved provider/owner to another.

Data analysis

Quality improvement was described by three categories: (1) centres that had No Change from Working Towards NQS; (2) centres that Improved to Meeting NQS; and (3) centres that Improved to Exceeding NQS. Our analytic approach compared these three groups in relation to the seven structural features (independent variables). First, we analysed the distribution of centres in each quality improvement category for each independent variable and their sub-categories (see Table 2). We then used multinomial logistic regression (MLR) analyses to test the significance of the observed differences in distribution across the three quality improvement categories. MLR is used in situations when the dependent variable has multiple, non-ordered categories. We selected the non-ordered logit model as the three categories may not have been equally spaced. The reference category was the mid-point, Improvement to Meeting NQS, with paired regressions to test Improvement to Meeting NQS versus No Change from Working Towards NQS and Improvement to Meeting NQS versus Improvement to Exceeding NQS.

The independent variables were all defined as categorical variables. Differences between the sub-categories for each independent variable were tested by comparing results for the first (reference) category to each of the other categories. For example, to test jurisdiction, we compared the distribution of centres across the three improvement groups for NSW to the distribution for each of the other states/territories. Reference categories for each independent variable can be seen in Table 3. We used two rounds of MLR tests: separate univariate MLRs to test the effect of each of the seven structural variables, and a multivariate MLR to test the combined effects of all seven variables. Multivariate MLR accounted for overlapping characteristics among the predictor variables in the model to identify the unique statistical effect of each structural variable.

The results of multinomial logistic regression tests are reported as unstandardised B coefficients, standard errors (SE) and Odds Ratios. The effect of each predictor variable is indicated by the direction of the B coefficient: a positive B coefficient indicates that the outcome being tested is more likely to occur for the comparison category than for the reference category; a negative B coefficient indicates that the outcome being tested is less likely to occur for the comparison category. The Odds Ratio describes the difference in the outcome for the test condition compared to the reference condition: an Odds Ratio of 1.00 indicates no difference; an Odds Ratio > 1 indicates the outcome is more likely to occur in the test category compared to the reference condition and an Odds Ratio < 1 indicates the outcome is less likely to occur in the test condition compared to the reference category. Analyses were conducted using STATA software, with significance set at p ≤ 0.05.


Descriptive statistics

Table 2 shows the distribution (number and per cent) of centres in each of the three groups (No Change from Working Towards NQS, Improvement to Meeting NQS, Improvement to Exceeding NQS) for each of the seven structural features and their sub-categories. Overall, 30.8% of long day care centres had no change from Working Towards NQS; 49.5% had improved to Meeting NQS; and 19.7% had improved to Exceeding NQS. The influence of each structural variable can be assessed by the extent to which these distributions vary within the sub-categories. For example, jurisdictional differences are suggested for Queensland and Victoria, which had lower proportions of centres in the No Change group (24.4% and 22.6% vs 30.8% for the No Change group overall), higher proportions in Improvement to Meeting NQS (62.8% and 64.9% vs 49.5% overall) and lower proportions for Improvement to Exceeding NQS (12.8% and 12.5% vs 19.7% overall).

Multinomial logistic regression—univariate

The results of seven univariate MLR tests confirmed a significant effect on quality improvement for all seven macro-structural features. For jurisdiction, improvement to Meeting NQS was more likely for centres in the Northern Territory (B = 0.73, SE = 0.34, p < 0.05), Queensland (B = 0.80, SE = 0.15, p < 0.05), South Australia (B = 0.63, SE = 0.29, p < 0.05) and Victoria (B = 0.91, SE = 0.17, p < 0.05) compared to NSW, and centres in Tasmania were more likely to improve to Exceeding NQS than centres in NSW (B = 0.97, SE = 0.46, p < 0.05). For location—degree of urbanisation, centres in inner regional areas were more likely to improve to Meeting NQS than centres in metropolitan areas (B = 0.31, SE = 0.15, p < 0.05). For location—community SES, centres in more advantaged communities (SEIFA 5/6, 7/8, 9/10) were more likely to improve to Exceeding NQS than centres in the most disadvantaged communities (SEIFA 1/2) (B = 0.61, SE = 0.21, B = 0.66, SE = 0.22, B = 0.79, SE = 0.21, respectively, ps < 0.01). For organisation type, centres operated by not-for-profit organisations were more likely to improve their NQS rating than for-profit, with significant results for Meeting NQS (B = 0.56, SE = 0.17 and B = 1.08, SE = 0.18, ps < 0.01) and Exceeding NQS (B = 0.85, SE = 0.20 and B = 1.81, SE = 0.20, ps < 0.01). Centres operated by large organisations were more likely to improve to Meeting NQS (B = 0.83, SE = 0.14, p < 0.01) or Exceeding NQS (B = 1.12, SE = 0.17, p < 0.01) than single, stand-alone centres. Similarly, at the centre level, larger centres with over 60 child places were more likely than smaller centres to improve to Meeting NQS (B = 0.50, SE = 0.11, p < 0.01) or Exceeding NQS (B = 0.74, SE = 0.13, p < 0.01). Centres that had a transfer of ownership were less likely to improve to Meeting NQS (B = − 0.25, SE = 0.11, p < 0.05) or Exceeding NQS (B = -0.73, SE = 0.14, p < 0.01) than centres that had a transfer of ownership.

Table 2 Distribution of long day care centres in improvement and no change groups by macro-structural features

Multinomial logistic regression—multivariate

Based on the results of the seven univariate MLR tests, we included all seven macro-structural features in the multivariable MLR tests. Table 3 presents sub-group comparisons for each variable for the paired outcomes: Improvement to Meeting NQS versus No Change from Working Towards NQS (columns 2–4) and Improvement to Meeting NQS versus Improvement to Exceeding NQS (columns 5–7). After controlling for the effects of all the variables in the model, six of the seven structural variables achieved significance for Improvement to Exceeding NQS, and three of these achieved significance for Improvement to Meeting NQS. Results are presented as a B coefficient, standard error (SE), Odds Ratio and p value (< 0.05 = significant) for each variable, as a contrast between the reference category and the other categories.

Table 3 Multivariate multinomial logistic regression comparing improvement to meeting NQS versus no change from working towards NQS and improvement to exceeding NQS


Odds Ratio results showed that centres in NSW were less likely to improve to Meeting NQS than centres in Queensland and Victoria (ORs = 0.54 and 0.50, ps < 0.01). On the other hand, Improvement to Exceeding NQS was less likely for centres in Queensland, South Australia and Victoria compared to NSW (ORs = 0.22, 0.31, 0.21, ps < 0.01).

Location—degree of urbanisation and community SES

When tested in the multivariate MLR, centres in metropolitan, regional and remote areas did not differ in their quality improvement outcomes, and only one of eight tests for community SES achieved significance. Centres in highly disadvantaged communities (SEIFA1/2) were less likely to improve to Exceeding NQS than centres in highly advantaged communities (SEIFA 9/10) (OR = 0.63, p < 0.00).

Type and size of provider organisation

Results showed that centres operated by for-profit providers were less likely to improve to Meeting NQS than not-for-profit centres, as evidenced by low Odds Ratios for the two groups of not-for-profit organisations: community-based organisations and school-based centres (OR = 0.53, p < 0.00); non-government and government operated centres (OR = 0.49, p < 0.00). A similar result was found for improvement to Exceeding NQS for centres operated by not-for-profit non-government and government organisations which were almost twice as likely to improve to Exceeding NQS as for-profit centres (OR = 1.94, p < 0.00).

Results for size of provider organisation showed that centres operated by small, stand-alone providers were less likely to improve to Meeting NQS than centres operated by large providers (OR = 0.52, p < 0.00). Equally, centres operated by large providers were more likely to improve to Exceeding NQS than those operated by small providers (OR = 1.49, p < 0.05).

Centre size and stability

Results at the centre level only achieved significance for improvement to Exceeding NQS. Larger centres were more likely to improve to Exceeding NQS than small centres (OR = 1.72, p ≤ 0.01), and centres that had experienced a transfer of ownership were less likely to improve to Exceeding NQS than those with stable ownership (OR = 0.68, p < 0.01).


This study of QRIS outcomes in Australia has shown that improvements in ECEC quality are influenced by a combination of macro-structural features, including state/territory jurisdiction, community SES, the type and size of approved provider organisations, and the size and stability of individual centres. In this section, we discuss these findings with reference to previous research and the unique context of Australian ECEC. It is important to note that the observed state/territory differences are likely due to underlying distinctions in ECEC policies prior to the major reforms of the national system in 2012 and the time needed for jurisdictions to implement these requirements. Controlling for jurisdiction effects was an important feature of our study design, which provides evidence of the impact of systemic, organisational and centre features on quality improvement, above and beyond the role of state/territory influences.

After controlling for other structural features, we found no effect for urban–rural location on QRIS outcomes and only a limited effect of community SES, suggesting that location and community disadvantage may be less influential on ECEC quality when other factors are controlled for. Community SES is known to affect the availability of ECEC, a factor highlighted by Hurley et al. (2022), who identified lower numbers of child places in areas of greater disadvantage, and by Cloney et al. (2016) who reported equi-proportional patterns of reduced provision in low SES areas for not-for-profit and for-profit services. In the present study, the four types of not-for-profit providers (community-based organisations and schools; not-for-profit organisations and governments) were more likely to have improved NQS A&R outcomes than for-profit providers. While ECEC usage is influenced by ECEC availability, the relationship between availability and quality remains unclear. Further research is needed to assess the interconnections among availability, type of provider and quality.

Our analyses further identified the independent effect of provider organisation size. Long day care centres operated by large multi-site organisations (eight or more services) were more likely to improve from Working Towards NQS to Exceeding NQS than single/stand-alone centres. In our view, these findings are likely due to larger organisations having economies of scale, which enable them to develop professional learning programs, curriculum support and resources to support staff. Importantly, our results showed that the positive effect of larger provider organisations occurred regardless of the governance structure of the organisation. We found a similar pattern for centre size: large centres (licenced for more than 60 children) were more likely to improve from Working Towards NQS to Exceeding NQS than small centres. Like larger providers, larger centres may have better access to resources, but their improvement in NQS ratings may also relate to the requirement for a higher proportion of degree-qualified teachers when the number of licenced places is higher than 60, and the associated benefits of a well-qualified workforce (Coley et al., 2016).

Centre stability, that is, not having a transfer of ownership during the period of time between QRIS assessments, was associated with improvement to Exceeding NQS. Given that staff stability, which provides certainty for families and trust for children (Heilala et al., 2022; Thorpe et al., 2020), is an important determinant of process quality, it may be that the effect of owner stability is associated with staff stability. Further research is needed to investigate the observed positive effect on quality improvement of stable ownership.

Limitations and implications

This study made use of administrative records of quality ratings gathered through six years of Australia’s national QRIS; however, unlike QRISs in the United States and some other countries, the NQS A&R measures are not based on the use of standardised observation instruments and, although conceptually sound (Jackson, 2015; Thorpe et al., 2021), only one validation of the NQS ratings has been conducted (Siraj et al., 2019). The lack of other validation studies, particularly for long day care centres, is a limitation of our study and, potentially, for any future research utilising these national datasets. Our study focussed on quality improvement in long day care centres with starting rating of Working Towards NQS. This restricted the sample to 1935 of the 3433 centres that had two or more A&R rounds. A further 176 centres had improved from Meeting to Exceeding NQS (see Table 1), which we did not include, but could potentially have added further insight into the structural predictors of quality improvement. Another limitation of the research is that the ACECQA dataset does not include information at the centre level about staff qualifications or staff–child ratios. While ECEC providers must meet legislated requirements for staffing, individual centres or approved provider organisations may choose to employ additional or higher qualified staff. Without this level of information in the data repository, it is difficult to fully explore the structural determinants of ECEC quality and quality improvement.

Despite these limitations, our analyses of Australia's QRIS data has identified predictors of improvement trends over time on a national scale. The results presented in this paper underline the need to address inequities between the for-profit and not-for-profit sectors and between services provided by small, stand-alone providers and large multi-site organisations. Given that 51% of Australia's ECEC services are for-profit and 80% are operated by stand-alone providers (ACECQA, 2022), our results suggest that greater resourcing is warranted to support centre owners/providers and staff to more effectively engage with QRIS processes. The similarity in improvement outcomes for centres within large, multi-site organisations and for large centres (> 60 places) signals the potential benefits of staff working in teams and creating collaborative organisational cultures, a feature identified as important for staff retention (McDonald et al., 2018). Further, this study has provided evidence that the stability of centre ownership is linked to quality improvement. Given the recent changes during the COVID-19 pandemic that has seen closures of centres and increased risks to staff' health and financial security (Logan et al., 2021), further research into the prevalence and impact of ECEC service stability is warranted. In sum, this study has underlined the important role of macro-structural influences, identifying predictors of quality provisioning and prioritising these for practice and policy reform.