Learner Models and SD Understanding

.Just presented two new research projects for the first time. Birgit Kopainsky (a million thanks!) presented at the System Dynamics Conference in Seoul while I took my chances in Freiburg at the E-Learning conference where I also had the chance to meet great people, especially Margot McNeill who one a best paper reward (well earned!). The Freiburg side of the conference organization was perfect, thanks to my colleague Dirk Ifenthaler, who also had a great paper there.

• Ifenthaler, D. (2010). The LMS moodle in schools: Benefits and obstacles. In M. B. Nunes & M. McPherson (Eds.), Proceedings of  the IADIS International Conference e-Learning (Vol. 1, pp. 81-88). Freiburg, Germany: IADIS.
• McNeill, M., & Diao, M. M. (2010). Student uses of IT in learning: An ethnographic study. In M. B. Nunes & M. McPherson (Eds.), Proceedings of  the IADIS International Conference e-Learning (Vol. 1, pp. 307-314). Freiburg, Germany: IADIS.
• Pirnay-Dummer, P. (2010). Theory-based case simulation and automated task synthesis to supoort learning on learning. In M. B. Nunes & M. McPherson (Eds.), Proceedings of  the IADIS International Conference e-Learning (Vol. 1, pp. 299-306). Freiburg, Germany: IADIS.
• Kopainsky, B., Pirnay-Dummer, P., & Alessi, S. M. (2010). Automated assessment of learners’ understanding in complex dynamic systems. Paper presented at the System Dynamics Conference in Seoul, South Korea, July 25 – 29, 2010.

Workshop on Theories and Models

Dr. Tristan E. Johnson, Dr. Dirk Ifenthaler and I just completed a wonderful workshop at Florida State University. The workshop was a modified version of those that were given in the years 2006, 2007, and 2008. This time it was about fundamental indeas about philosophy of science, theory bulding, theory quality and testing and the implications for reserach designs and methodology. For all parts, best practice examples were provided. Also, we embedded tasks for all stages of the research process so that the doctoral students could apply the contents directly to their own research interests. We even had guest speaker special appearances of Dr. Val Shute and Dr. Gershon Tenenbaum.

Combine Scales in R

Good instruments for cognitive — and thus not directly oberservable — constructs  measure indirect properties with more than one item. If the items are consise (high scale reliability, e.g. conbach’s alpha), then they are combined. Some of them by sums and means, others by more elaborate functions depending on how the single items are related to each other. Different methods are usually available to calculate the scale value. Some are even still on paper. Of course, it is easy to combine the scales in Excel or Open Office. However, if there is more than one study involved, one might want to automate things a bit more — and not manually generate those functions over and over again. R provides the necessary means, but they are not very intuitive. Let’s say, we have a simple scale in the data.frame D with the variables (names) a,b,c and d. The scale is combined with a simple scale mean.

The following example would work, but it is not very nice — especially when the scale contains more items.

D$scale = (D$a + D$b + D$c + D$d)/4

This would work, because all elements are treated as vectors. The following would however not work because the mean is calculated with the whole data and not the row.

D$scale = mean(c(D$a, D$b, D$c, D$d), na.rm=TRUE)

The solution is a workaround which uses a matrix and it works for any function.

D$scale <- apply (matrix(c(D$a, D$b, D$c, D$d), ncol=4), 1, mean, na.rm=TRUE)

The reason for referring to every variable individually is a practical one. Variables of a scale are rarely located next to each other in a test. ncol refers to the number of columns (wie do have 4 variables here). The 1 instructs the apply-function to calculate the means per row (2 would be per column, which wouldn’t make much sense here). na.rm=TRUE instructs the mean function to remove missing values rather than to skip the whole set if a single value is missing. Whether I’m allowed to do that depends on the instrument.

Links

• Apply
• Data.frame
• Matrix
• Mean
• Comprehensive R Network

CELDA 2009 in rome

CELDA 2009 in Rome was a great success. The conference had an acceptance rate around 20% and was able to raise its quality while also bringing more participants. The organizers (among them my valued colleage Dirk Ifenthaler)  did a wonderful job in creating a collaborative and creative atmosphere throughout the participants who came from 37 different countries and from all over the world. Beside the good meetings with cooperation partners, I also was on 3 paper presentations. Thus I did only have little time left to explore the city of Rome (will definitly come back).

Pirnay-Dummer, P., Ifenthaler, D., & Rohde, J. (2009). Text-guided automated self-assessment. In Kinshuk, D. G. Sampson, J. M. Spector, P. Isaías & D. Ifenthaler (Eds.), Proceedings of the Celda 2009, cognition and exploratory learning in the digital age, 20-22 November (pp. 111-116). Rome, Italy: Iadis.

Rauh, K., Pirnay-Dummer, P., & Roeder, A. (2009). The effect of writing style on text-based knowledge externalizations made using T-MITOCAR. In Kinshuk, D. G. Sampson, J. M. Spector, P. Isaías & D. Ifenthaler (Eds.), Proceedings of the Celda 2009, cognition and exploratory learning in the digital age, 20-22 November (pp. 93-100). Rome, Italy: Iadis.

Wijekumar, K., & Pirnay-Dummer, P. (2009). Assessing diversity disposition of European and American students. In Kinshuk, D. G. Sampson, J. M. Spector, P. Isaías & D. Ifenthaler (Eds.), Proceedings of the Celda 2009, cognition and exploratory learning in the digital age, 20-22 November (pp. 519-520). Rome, Italy: Iadis.

Theory and Practice

Today, we discussed the area of possible conflict between theory and practice. Coming from fundamental research I favor the position that a decision based on sound theory is always prefarable over a decision on single cases. The risk of error is lower in that way. Also, finding the cause for effects is easier with a good theoretical foundation. However, two important requirements are often ignored:

1. There actually has to be a set of assumptions that fulfills the requirements of a theory (a set of somewhat thought over ideas is not a theory)
2. The theory needs to serve as a prerequisite. Very often pratictioners develop something nice and look for a fitting theory afterwards. Most of them end up explaining why the application did not work.

Suprisingly, well formed theories are still rare in the field of learning and instruction. And even researchers often confuse the term “theory” with “opinion” or “idea”. The field desperatly needs a good formulation of theories, explicitly stating the axioms and theorems and deductivly develop hypotheses which will have to be empirically tested.

‘Interconnecting Research to Practice: Is there any real benefit or is it mostly bother?’ Add to my schedule   Description:   The main goal for this interdisciplinary panel is to engaging in a dialogue between researchers and practitioners of educational and instructional solutions in order to create and share an understanding of the roles of research in practice and practice in research. We will discuss the benefits and cost of exploring research and practices. The panel will explore whether research and practice compliment one another, what are the challenges of merging research and practice, and what are successful moments that panelists have experienced merging research into practice or practice into research.

AECT 2009. Louisville, Kentucky – Integrative Panel
Key presenter:   Tristan Johnson, Florida State University Copresenter(s):   Lisa Yamagata-Lynch, Northern Illinois University | Dirk Ifenthaler, University of Freiburg | Sharon Smaldino, Northern Illinois University | Dan Schuch, Pacificorp Corp | Andrew Gibbons, Brigham Young University | Val Schutes, Florida State Univeristy | Lara Luetkehans,  | Simon Hooper, University of Minnisota | Pablo Pirnay-Dummer, University of Freiburg | Marci Murowski