Our MethodologyOur Methodology Angela Ball Wed, 08/19/2020 - 14:55
Insights from JEDI
The NSC uses insights from JEDI, or the Jobs and Education Data Infrastructure project.
JEDI pioneers a new approach to skills-based labour market analysis. It does this by defining skills as the common language linking jobs to education and training. By combining traditional and near real time data using data science techniques, JEDI can identify transferable skills and how skills are changing in the labour market.
Guiding principlesGuiding principles Angela Ball Wed, 08/19/2020 - 14:57
In order to identify the emerging occupations, four guiding principles were followed:
- Data driven: the NSC used Burning Glass Technologies data, O*NET (the American occupation classification) and ANZSCO to identify emerging occupations. The NSC then validated and created profiles for these emerging occupations using microdata from the ABS Labour Force survey and the ABS Employee Earnings and Hours survey.
- ANZSCO based: the emerging occupations align with the ABS concept of an occupation and reflect occupations that are not currently part of ANZSCO.
- Critical mass: the emerging occupations must occur frequently enough in job advertisements to be classified as new occupations (at least 100 job advertisements over the last 5 years).
- Substantially different: the emerging occupations must not be alternate titles of existing occupations or a ‘strict subset’ of existing occupations. For example, a cyber‑security expert shares enough similar tasks with an ICT security specialist that we consider this is an alternate title for this occupation, rather than a substantially different emerging occupation.
Complementary methodsComplementary methods Angela Ball Wed, 08/19/2020 - 14:58
No one method is sufficient to identify all emerging occupations.
The NSC has taken three complementary approaches (Figure 3), which involve:
- Referencing skills projections produced by Burning Glass Technologies based on internet job advertisements. The NSC identified top job titles associated with high-growth skills and manually reviewed the job titles for these to identify genuinely new roles. As the top growing skills were often technology tools, many of the top growing ‘job titles’ reflected that tool (for example, .Net Developer). Such titles were excluded from the analysis.
- Analysing the linkages between Australian job advertisement data from Burning Glass Technologies and other classification systems (such as O*NET), to give an indication of acceptance of emerging occupations in other contexts. This was followed by quality assurance processes to ensure these occupations align with the guiding principles of the project.
- Reviewing job advertisement data from Burning Glass Technologies to identify job titles that have at least doubled over the last five years. While this method was the most qualitative of the approaches, job titles are sometimes used by employers to signal a substantially new job is emerging, so we conduct this exercise for comprehensiveness.
Figure 3: Emerging occupations identified by the National Skills Commission
Validating emerging occupationsValidating emerging occupations Angela Ball Wed, 08/19/2020 - 14:59
To validate emerging occupations, the NSC used unit record data from the ABS quarterly Labour Force surveys (from 2014 to 2019) to ensure that the proposed emerging occupations are understood by people in the labour market and are substantially different from existing ANZSCO occupations.
The NSC considered the survey’s text fields for occupation title, task and industry, in combination with text mining techniques to search for titles, alternative titles and skills. The outcome was then qualitatively reviewed to ensure that the data sample described the occupation properly.
Text mining and statistical analysis of the Labour Force microdata (2014 to 2019) and Employee Earnings and Hours microdata (2018) outputs were used to create occupation profiles for each emerging occupation, showing employment size and other demographic characteristics. Burning Glass Technologies job advertisement data was used to determine in-demand skills. In order to produce reliable and representative occupation profiles, we required the data sample size to be at least 30 unit records for ABS microdata, and at least 100 job ads for Burning Glass Technologies data. We also used a two-year moving average method to smooth employment size data over time.