Getting the funding right
This report was originally published 19 August 2013.
What is FOEI?
FOEI is a school socio-economic index that is based on parents’ highest level of school education, non-school qualification and occupation status. This information is captured on school enrolment forms and recorded in the Enrolment Registration Number (ERN) system. Data is extracted in early Term 2 each year.
FOEI includes all students, enrolled in all NSW government schools, including preschool students and Schools for Specific Purposes (SSP).
FOEI uses a statistical method to weight and combine parent information into an index that represents each school’s average socio-economic disadvantage relative to other NSW government schools.
For many years governments have tried to address the equity problem that students coming from poorer backgrounds are likely to be educationally disadvantaged and have lower levels of achievement.
The Centre for Education Statistics and Evaluation (CESE) has developed a new measure of school socio-economic status (SES), the Family Occupation and Education Index (FOEI), to accurately identify levels of socio-economic disadvantage.
Why do we need FOEI?
The new national Schooling Resource Standard used in the National Education Reform Agreement (NERA) includes needs-based funding or ‘loadings’ for all students from the most disadvantaged half of the population. The rate of funding is determined by the concentration of disadvantage in each school.
Under NERA, each state government is required to introduce its own needs-based funding arrangements that are consistent with the federal model, and to publish both the actual funding allocations and how they have been calculated.
During 2012-13, the Department of Education and Communities (DEC) has been developing a new Resource Allocation Model (RAM) which includes a loading for students from socio-economically disadvantaged backgrounds. FOEI provides the measure of SES disadvantage for this loading.
What does FOEI look like?
FOEI is a score ranging from 0 to approximately 300, with a mean of 100 and a standard deviation of 50. Higher FOEI scores indicate higher levels of need (i.e. lower SES.)
In addition to the overall school FOEI score, the distribution of students across FOEI quarters is also used in the resource allocation model. To generate each school’s quarter distribution, an SES score is calculated for each student from a combination of parent education and occupation information. From the DEC-wide distribution of students’
scores, the cut-points for the 25th, 50th and 75th percentiles are located and used to assign each student to a quarter. For each school, the percentage of students in each quarter can then be determined.
What is FOEI composed of and why?
FOEI has been in development since 2009. It was developed as a measure of SES, which for children and young people is typically measured in the research literature by three core, inter-related components (eg. Butler, 2012; and Marks et al., 2000):
- level of parental education,
- parental occupational status, and
- family/parental wealth.
Analysis undertaken by DEC confirmed previous research findings that parental education attainment is one of the strongest predictors of student and school performance. If parental occupation is added to this, the predictive power is further enhanced. Collectively, parental education level and occupation status accounts for more than 70 per cent of the variation in performance across schools.
How do we know that FOEI is accurate?
FOEI has been validated against other measures including the Priority Schools Funding Program (PSFP) index and the Index of Community Socio-Educational Advantage (ICSEA).
If FOEI is adjusted to include similar components as PSFP and ICSEA (i.e. Aboriginal background), it has greater predictive power for school performance than either of the other two measures.
There is also a high level of agreement between school-level FOEI values and the distribution of students across FOEI quarters.The National Institute for Applied Statistical Research Australia at the University of Wollongong has independently reviewed and validated the process used to obtain the FOEI.
Refinements to FOEI in 2013 make it the best available SES measure, as a basis to allocate resources to NSW government schools enrolling students from low SES backgrounds. FOEI is stable for most schools from year to year, with variations clearly relating to changes in the underlying data.
FOEI will be subject to a process of continuous improvement. This will involve an annual review of the methodology, and continued efforts to work with schools and principal networks to improve the completeness and quality of data recorded in the enrolment system.
Why can’t we use existing measures of school disadvantage?
FOEI is a better and more responsive measure than ICSEA or PSFP for funding NSW government schools with students from socio-economically disadvantaged backgrounds, as Table 1 shows.
FOEI is based on an annual extract of data from the enrolment system, so it captures information about the latest student population enrolled in each school.
ICSEA will always be two years out of date if used in funding for NSW government schools. Funding calculations for a given year (e.g., 2014) begin in the year prior (e.g., 2013), however at that point in time the only ICSEA data available is from the previous year (e.g., 2012).
The PSFP survey was only conducted every four years.
FOEI is based only on NSW government school students, resulting in an index that is specific to NSW government schools educational contexts.
ICSEA is a national measure and includes both government and non-government schools, so the distribution of ICSEA values and student quarters reflects all schools and students across Australia. For example, only around 20 per cent of NSW government students are in the bottom ICSEA quarter nationally. However, the state distribution of low SES funding needs to apply the greatest loadings to the bottom 25 per cent of students in NSW government schools.
The PSFP index is only relevant for approximately 50 per cent of schools.
FOEI can be calculated for all NSW government schools.
ICSEA values and quarters are not available for all schools. On the My School website, around 140 NSW government schools are missing ICSEA values and around 180 are missing ICSEA quarter data.
Data is only available for schools that participated in the PSFP survey.
FOEI is based only on core socio- economic background factors. (Note that RAM already includes funding components to address the educational disadvantage associated with other factors such as Aboriginal background and remoteness.)
ICSEA is based on factors additional to core socio-economic background factors. As ICSEA is used for comparing and interpreting school performance it includes additional student background factors (i.e. Aboriginal background, remoteness) that relate to performance.
The PSFP index also includes additional factors such as Aboriginal background.
FOEI uses a robust regression technique to manage outliers (i.e. schools that do not fit the model as they exhibit a different pattern of relationship between parental background factors and school performance), resulting in a more stable model for FOEI values.
ICSEA is based on a regression technique that is susceptible to outliers. This contributes to year- to-year variation in ICSEA values for all schools.
The PSFP index also uses a regression technique that
is susceptible to outliers.
FOEI uses direct student enrolment data and has an established and principled method to reduce bias due to missing data.
ICSEA deals with missing parental data by either ignoring it or by constructing ICSEA from a different set of variables (i.e., ‘community level data’ from the ABS census data linked to student addresses). For 2012 ICSEA values reported in My School in 2013, 286 (14 per cent) schools had ICSEA values based on census data rather than direct parent data. Therefore ICSEA values are not strictly comparable across all schools.
The PSFP index is based only on a sample of parents for most schools. Missing data rates were low, but when parental data was missing it was ignored.
FOEI has been developed, and will continue to be supported by CESE.
Reliant on a third party to develop, support and disseminate a measure (although currently free-of-charge.)
The PSFP survey is very costly and a burden on schools.
FOEI is a high quality index that addresses statistical problems effectively.
Using established statistical methods to reduce bias arising from missing data
FOEI uses an established statistical methodology to impute values for missing data. This methodology generates multiple plausible values for the missing data based on the observed relationships between parental variables and other related variables for students where data is available. In total, 10 ‘complete’ datasets are produced, each consisting of the existing data and one set of plausible values for the missing data.
For each school, the percentage of parents in each education and occupation category is calculated for each ‘complete’ dataset and then averaged across the datasets for use in the FOEI calculation. Similarly, the distribution of students across the quarters is calculated for each dataset and then averaged to determine the final quarter distribution.
The overall impact of the imputation method on the distribution of parental variables is that more parents are in the low SES categories, based on post-imputation data compared to the observed data. This indicates that the imputation has helped to produce a more accurate picture of the SES make-up of the parental population in NSW government schools, given the observed pattern of lower SES parents being less likely to provide some background information on student enrolment forms than parents of higher SES backgrounds.
Using a robust modelling technique to reduce the undue influence of outliers
CESE analysis shows that the majority of outliers are small schools and selective schools. A few outliers can result in a model that does not best reflect the majority of schools.
CESE’s examination of alternative regression techniques showed that the best way to reduce the influence of these outliers and improve the stability of the FOEI model from year to year, is to use a technique known as robust regression.
This technique, used to generate the model for calculating FOEI values, allows all schools to contribute to the overall model as a function of how well they fit the predominant pattern.
Additional data treatment practices further strengthen the quality of FOEI
FOEI for SSPs is based on all students
To better reflect the socio-economic background of the students attending SSPs, parental background information was used for all students enrolled in or attending SSPs (i.e. both ‘census’ and ’non-census’ enrolments). This increased the number of students upon which the FOEI calculation is based, leading to greater accuracy and stability of FOEI for these schools.
Equal weighting for students from different family types
In other measures, information for students in single-parent families has been under-represented relative to students with information for two parents. FOEI corrects for this by ensuring that each student’s family background information is given the same weight.
FOEI is the best SES measure for use as the basis to allocate resources to NSW government schools which enrol students from low SES backgrounds. It is a fairer and more accurate measure than existing measures (PSFP and ICSEA) for identifying relative levels of socio-economic disadvantage among NSW government schools.
More information about the FOEI methodology and analysis results can be found in the technical report.