E-Research for Education

This entry is an eclectic summary of the key trends in scientific research methodologies, technologies and practices followed by some reflections about the state of the art and future of the educational research. Essentially this blog is a mashup of ideas from three unrelated in a structured world readings and some outsider's thoughts that link them in a complex world.

Sources:

• Alex Szalay: Science in an Exponential World. Paper presented at eResearch Australasia Conference, Brisbane, 26-29 June 2007. URL
• OECD: Evidence in Education: Linking Research and Policy, 12/06/2007. OECD, CERI. URL
• Uri Wilensky and Michael J. Jacobson: Complex Systems in Education: Scientific and Educational Importance and Implications for the Learning Sciences. Journal of the Learning Sciences. 2006, Vol. 15, No. 1, Pages 11-34. URL

Evolution of scientific inquiry

1. EMPIRICAL: Thousands years ago in Aristotle’s era modern science was empirical, i.e. based on careful observation and description of natural phenomena.
2. THEORETICAL/LOGICAL: Hundreds years ago modern science became theoretical. E.g. Newtown’s, Kepler’s, Euler’s and other similar discoveries were based on logical reasoning, generalisations and mathematical manipulations.
3. COMPUTATIONAL: About two decades ago science became computational, i.e. based on computational modelling, simulations of complex phenomena, visualisations, etc.
4. EXPLORATORY and DATA DRIVEN: Today’s main scientific discoveries are based on data exploration, i.e. synthesis of theory, experiment and computation using advanced data management and statistics (e.g. data mining).

View from the educational trench: Almost all present educational research belong either to empirical or to logical research category (i.e. “qualitative” and “quantitative”). Computational and data driven research are rather vision (e.g. Wilensky & Jacobson, 2006) and isolated experiments (e.g. DM website) than valued evidence for educational decision making (e.g. OECD, 2007).

Critical educational issues in essence require complex investigations of social and educational processes (not only outputs) and modelling (i.e., computational and exploratory techniques).

Tested rules of thumb for scientific discoveries

• Sensors for data gathering become inexpensive
• Federation of N data repositories has utility O(N²) and possibilities for new discoveries rise at ratio O(N²)!
• Discoveries are made…
  • at the edges, boundaries, inter-connections.
  • going deeper, collecting more data, using better visualisations.

View from the educational trench: Some “sensors” are cheap (e.g., e-learning transcripts, administrational data), some are still expensive (e.g., national and international surveys). Most present educational datasets are small, distributed and not interrelated. Most datasets typically belong to a single researcher, group or institution. They are rarely reused and typically discarded at the end of the research project. Data valued by its users (i.e. individual researchers), but not society.

Research of lifelong ubiquitous learning needs continuous integrated data flow from “multiple” sensors (i.e., quantitative and qualitative in the form of numbers, texts, sound, video…) during entire lifespan. Present data collection, management and preservation approaches create very few opportunities to go deeper, integrate or look at boundaries.

Sociology and technology

• Technology driven social changes are unpredictable, and nobody knows what will come and/or go next: YouTube, Google?
• Technologies (and world) are changing too fast: there is no time for top-down engineering of networks, data and/or research.

View from the educational trench: (E-)educational developments partly mirror technology-driven social developments: e-learning, m-learning, W2-learning, IPod-learning, SecondLife-learning. Some of them will soon be discarded, some will stay and some new will come next. There is no time for top-down engineering society or education systems. Typical strategic planning of educational innovations and research becomes impossible and inappropriate.

A possible research model is highly distributed and very close to “doers” (learners/teachers) educational research. In this scenario, the first challenge will be to federate and integrate autonomously collected datasets. The next challenge will be to find socially desirable ways to make such (socially sensible) datasets easily available for new discoveries.

Emerging paradigm of the scientific research

• “Optimal” statistics have poor scaling: for large datasets major errors become statistically insignificant.
• Idea of “optimal” should be discarded: world is complex, data is fuzzy and answers will be always approximate.
• Researchers must take risks: it is impossible to get it always right.

View from the educational trench: Discourses from the recent OECD (2007) report about the trends in evidence-informed educational policy:

• “… experimental designs, and especially that of randomised control trials, should be given a stronger role…” (p. 9).
• “… the research that is available is contradictory and does not suggest a single (sic!) course of action that could be reflected in policy” (p. 9).
• “The importance of deciding first on a research question and then choosing the appropriate methodology with which to investigate the question is clear (Shavelson and Towne, 2002), yet often this point is overlooked by researchers and policy makers alike (Berliner, 2002).” (pp. 23-24).

Although it is well realised that: “…many research products are presented much too late to fit into the needs of modern politics. They are too slow afterthoughts than vital challenges.” (p. 149).

Emerging paradigm of the educational research?

Last view from the educational trench: Educational research should go beyond typical “qualitative”, “quantitative” or “mixed method”, to include computational and exploratory data research methods, federated datasets, collaborative research culture and new ways for collecting data and discovering knowledge. Educational research needs a new pervasive lifelong learning research paradigm. This new research paradigm should be based on different epistemic assumptions about learning and construction of scientific evidence - pervasive, ubiquitous and ambient learning and research assumptions.

Dream or vision? - Risk.

Acknowledgements: Most facts and ideas about the scientific research are based by Alex’s Szalay’s presentation. All (mis)interpretations are LM’s responsibility.