How "open" are universities? An Australian case study

The higher education sector is rapidly under pressure to evolve. One driving force is the commitment to “openness” in knowledge production and dissemination. While changes for open science and open access (OA) have picked up momentum in recent years, there remains much unknown in other areas of openness such as wider community engagements and inclusivity. “Open knowledge institutions (OKIs)” is a framework that advocates for universities to operate with principles of openness at their centre; not only should their research be made accessible, but they should lead positive changes to diversity, equity, transparency, etc.

The Curtin Open Knowledge Initiative (COKI) is a project that sets out to explore different facets of OKIs. One of its aims is to build tools in supporting evidence-based strategies for universities to progress as OKIs. As such, COKI has collected and integrated data from a number of sources that relates to openness, including information on OA publications, research collaboration, social media activities, diversity, general online and physical accessibility; and, related aspirations and actions signaled through policies, infrastructure and annual reports.

This presentation reports on the data and exploratory analysis for Australian universities focusing on the year 2016 as a case study. This includes 43 Australian universities and 26 potential indicators for OKIs (plus an indicator on revenue). Initial descriptive analysis shows that many of these indicators are characterised by extreme observations, in addition to skewness, missing values and largely different measurement scales. These make it necessary to implement robust statistical methods, such as the use of Spearman’s rank correlation, robust principal component analysis (PCA) and robust clustering methods.

Our analysis shows that the OKI indicators can be largely characterised into three groups of “diversity”, “communication” and “coordination”, which closely resembles the OKI evaluation framework proposed by Montgomery et al. (2018). We also show that the information provided by these indicators are high-dimensional and complex.

The results also show evidence of:

• large disparities in indicators such as indigenous employment and gender equity.

• a strong negative correlation between diversity and physical accessibility.

• traditionally higher ranked universities (in international rankings) performing relatively worse in diversity indicators.

• a preference for repository-mediated OA.

• low correlation between university revenue and OA publishing; and, lower ranked universities do not necessarily perform worse in terms of proportion of OA outputs.

• synergies across universities within the same university network, which is most prominent for the Group of Eight.

Although the study draws on data focused on Australian universities, many of the findings can be related to wider literature on universities. For example, there is clear correspondence to the general consensus that advances in gender diversity remain less in senior academic positions, and the probable links between OA performance and local policy strengths. The work also provides insights into challenges of data collection, integration, and analysis; and, a number of potential links between different aspects of university openness. These lay foundations for further large-scale and longitudinal studies.