Appendices
A.1 Guide to interpreting figures and tables in this report
Percentiles and box plots
This report uses box plots to visualise the position of the 5th, 25th, 50th, 75th and 95th percentiles for the Base Salary, Total Remuneration Package and Total Reward. In this report percentiles mark intervals where data occurs starting from given points in the entire dataset. Note that the 25th and 75th percentiles are referred to as Q1 and Q3 respectively. These intervals are described in the table below.
|
Percentile point |
Also known as |
Percentage of data below point |
Percentage of data above point |
|---|---|---|---|
|
5th |
P5 |
5% |
95% |
|
25th |
Q1—First quartile |
25% |
75% |
|
50th |
Median |
50% |
50% |
|
75th |
Q3—Third quartile |
75% |
25% |
|
95th |
P95 |
95% |
5% |
Example plot elements
The following example shows box plots of base salaries paid to employees in Classification 1 in two consecutive years, A (left column) and B (right column). The vertical axis represents the amount of money paid. The percentiles are represented by horizontal lines and are labelled at their appropriate positions. The median is represented by the middle line of each set of 4 boxes for the two years.
The box colours are different to distinguish between intervals and focus your eye on the spread of data from the first to third quartile, where the majority of data lies. They have no other meaning.
A larger box between percentiles indicates a greater range between the largest and smallest salaries paid in that interval. A smaller box indicates a smaller range. Therefore, in year B, the pay range increased at Q3—shown by larger distances between the median and Q3 relative to year A. In addition, there was no change in the 5th percentile, Q1 and median from year A to year B.
Example plot of percentile
Methods
All APS agencies were required to report data for all employees that were employed under s22(a), s22(b) and s72 of the Public Service Act 1999 as at 31 December 2018.
Raw data was collected from agencies using standard guidelines set out in the APSC Data Specifications. Submissions were checked by the APSC against the APSED database, agency pay points (from Enterprise Agreements), and approved, clean data received by agencies in the previous year.
Data that failed the APSC quality checks were returned to agencies for correction and resubmission, after which the data was returned for final clearance and incorporation into the remuneration dataset. Agencies are required to sign off on the accuracy of their data prior to inclusion in the report.
The following steps were taken to standardise the collected data. For part-time employees, data was recalculated into the full time equivalent (FTE). Figures were annualised for employees who worked for only part of the year (though active as at 31 December 2018).
Employees who were in a graduate program in 2018 were categorised as ‘Graduates’, even if they had advanced to other classifications by 31 December 2018. Remuneration for these employees was recorded as at their last day as graduates. An anomaly in the application of this rule in 2017 graduate data was identified during the development of the 2018 report. The 2017 data set has been re-based to correct this.
The weighted average median (WAM) is used throughout the report to provide an estimate of percentage changes in remuneration over time while accounting for changes in headcount at each classification. This is an historical measure from successive APS remuneration reports.
Some columns in the tables may not add up because TRP and TR are calculated separately for each employee. These are the values which determine the median. Therefore median TR is not always the sum of all medians for Base Salary, TRP and TR.
Data from this report should not be used to calculate past or present populations of the APS. For accurate data as at 31 December 2018, please refer to the 31 December 2018 Employment Data Release, which is available from the APSC website. Analysis comparing data reported prior to 2017 against the current report has identified discrepancies with some of the figures following a change in the software used by the APSC for reporting.
Note that the data reported for previous years as presented in the current edition of the APS Remuneration Report will differ from reports published before 2017.
The APSC has started publishing the APS Remuneration Report in Tableau, having previously used SAS. This has resulted in some changes to how the reported distribution data is calculated. This is because the two programs use different, patented methods for calculating percentiles when even numbers of observations occur. Such differences in calculation have had a minor impact on the position of percentiles as reported in previous years.