Donnerstag, 30. Mrz 2017, 17:00 - 19:00 iCal
Lessons learned (from mistakes)
Part 1: Foundations of response time measurement
Part 2: Buggy rules in quantitative reasoning
Fakultät für Psychologie, Hörsaal G
Liebiggasse 5, 1010 Wien
Vortrag
Wir laden Sie herzlich zum Gastvortrag von Prof. Dr. Matthias Gondan–Rochon am 30. März 2017 ein und freuen uns auf Ihr Kommen.
Part 1: Foundations of response time measurement
Conclusions on mental organization and performance are often based on the time it takes for a correct response to a given problem. However, participants do not always demonstrate 100% accuracy, and the response time distribution is contaminated by outliers, omissions and mistakes. Most RT analyses in Cognitive Psychology exclude all problematic responses (“mean correct response time”), although a wrong response after some t carries information, namely that the participant was unaware of the correct response until t. I will present imputation techniques that use this information. Using patient data and simulations I will demonstrate that reliability and validity of response speed estimates can be improved when the mistakes made by the participants are taken into account (e.g., Koch et al., 2013).
Part 2: Buggy rules in quantitative reasoning
I present an E-Learning tool that anticipates and detects buggy reasoning in quantitative problem solving. The can then provide feedback for specific mistakes that underlie the erroneous solutions. For example, some students forget the (implicit) parentheses around the numerator and denominator of fractions, so that a fraction like X – ? in the numerator and s / ?N in the denominator is incorrectly calculated as X – (? / s) / ?N because “X – ? / s / ?N” is linearly typed into the calculator. More abstract buggy rules include the use of irrelevant variables as predictors in regression models, wrong statistical tests, or inadequate handling of missing data. Such mistakes follow their own logic, so that they can be implemented as Prolog rules (Colmerauer & Roussel, 1996), combined and recognized by a computer system (e.g., Zinn, 2006). By extending the elementary algebraic functions as well as typical statistical procedures, the system can diagnose a potentially huge set of students’ answers. The Prolog interpreter is embedded as an R library (R Core System, 2017; rolog) which enables the simultaneous use of a declarative and imperative programming language.
Veranstalter
Prof. Dr. Helmut Leder, Dr. Michael Forster
Kontakt
Abla Marie-Jose Bedi
Fakultät für Psychologie
Institut für Psychologische Grundlagenforschung und Forschungsmethoden
+43-(0)1-4277-471 04
abla.bedi@univie.ac.at
Erstellt am Dienstag, 21. Mrz 2017, 14:45
Letzte Änderung am Mittwoch, 22. Mrz 2017, 10:34