Effect of Student Ability and Question Difficulty on Duration

Time has become a standard feature used in EDM models, and is used in models of meta-cognitive strategies to models of disengagement. Most of these models consider whether a student action is "too fast" or "too slow". However, an open question remains on how we define and select...

Full description

Saved in:
Bibliographic Details
Published inInternational Educational Data Mining Society
Main Authors Ma, Yijun, Agnihotri, Lalitha, Baker, Ryan, Mojarad, Shirin
Format Report
LanguageEnglish
Published International Educational Data Mining Society 2016
Subjects
Academic Ability
Academic Achievement
Difficulty Level
Electronic Learning
Item Response Theory
Knowledge Level
Reaction Time
Student Reaction
Test Items
Online AccessGet full text

Cover

More Information
Summary:Time has become a standard feature used in EDM models, and is used in models of meta-cognitive strategies to models of disengagement. Most of these models consider whether a student action is "too fast" or "too slow". However, an open question remains on how we define and select these cut-offs. Moreover, it is not clear that the same cut-offs are appropriate across different situations. Some students may generally respond faster than others; more difficult items may take different amounts of time. In this paper, we consider whether absolute or relative indicators of time are more appropriate as cut-offs, and whether simple transformations (such as log time) are useful when representing time. We do so through visualizing student performance in relation to general student ability, item difficulty, and different ways of representing time. We find that student knowledge and item difficulty should be taken into account when choosing cut-offs, and that there are advantages to representing duration in terms of standardized log-time. [For the full proceedings, see ED592609.]