sequential and temporal data jonna malmberg. what can sequential and temporal data reveal?
TRANSCRIPT
SEQUENTIAL AND TEMPORAL DATA
Jonna Malmberg
WHAT CAN SEQUENTIAL AND TEMPORAL DATA REVEAL?
WHAT IS SEQUENTIAL AND TEMPORAL DATA?
• Data about the (learning) process• Quite complete (relative to what it is
possible to observe)• Very fine-grained, time-stamped
representation• Video data, eye movement data, log file traces,
chat data
– Unobtrusive – data are created as learners do what they do
Perry & Winne, 2011
WHY TEMPORAL AND SEQUENTIAL DATA
• The way in which students engage in self-regulated learning (and SSRL) is affected by previous learning experiences and contextual, situation-specific features (Azevedo et al, 2010).
- The events are not independent (Hesse, 2013)
• Analysing temporal sequences of SRL and SSRL informs regulatory processes as they unfold (Järvelä & Fisher, 2014)
• How to identify temporal sequences of SRL and SSRL?
DATA EXAMPLE Case study from higher education context
18 graduate students worked in 6 groups over an 8-week in a “Learning for Understanding” course
• The course involved three learning phases, each focusing on a specific topics. The collaborative tasks were constructed to be challenging and require multiple student perspectives.
Cognitive, Motivational and Emotional Base for Learning for Understanding Task 1 Task 2 Task 3
Learning sciences and Self-regulated learning
Memory structures and Learning strategies
Motivation regulation and Instructional principles
nSTUDY LEARNING ENVIRONMENT (Winne, Hadwin & Beaudoin, 2010)
Support for self and shared regulation of learning• Planning – and reflection notes• Chat discussions
Data sources: Recorded• log file traces • chat discussions
LOG FILE TRACES
• Example of Logfile traces from nStudy learning environment (Winne et al., 2011)
OpenWindowAction 1.32205E+12 2011-11-23 05:26:01GotFocusAction 1.32205E+12 2011-11-23 05:26:07BrowserDocumentOpened 1.32205E+12 2011-11-23 05:26:08LostFocusAction 1.32205E+12 2011-11-23 05:26:09BrowserDocumentOpened 1.32205E+12 2011-11-23 05:27:43GotFocusAction 1.32205E+12 2011-11-23 05:27:46LostFocusAction 1.32205E+12 2011-11-23 05:27:46OpenWindowAction 1.32205E+12 2011-11-23 05:27:46ClickButtonAction 1.32205E+12 2011-11-23 05:27:46GotFocusAction 1.32205E+12 2011-11-23 05:27:50FolderSelected 1.32205E+12 2011-11-23 05:28:29FolderSelected 1.32205E+12 2011-11-23 05:28:32FolderSelected 1.32205E+12 2011-11-23 05:28:33FolderSelected 1.32205E+12 2011-11-23 05:28:36ItemSelected 1.32205E+12 2011-11-23 05:28:38NoteWindow.Save1.32205E+12 2011-11-23 05:28:38LostFocusAction 1.32205E+12 2011-11-23 05:28:38OpenWindowAction 1.32205E+12 2011-11-23 05:28:39GotFocusAction 1.32205E+12 2011-11-23 05:28:40BrowserDocumentOpened 1.32205E+12 2011-11-23 05:28:40LostFocusAction 1.32205E+12 2011-11-23 05:29:20BrowserDocumentOpened 1.32205E+12 2011-11-23 05:29:22BrowserDocumentOpened 1.32205E+12 2011-11-23 05:29:26
• Usually thousands of rows of information
• Not all the information is important
View events = When something is opened Model events = When something is a) Savedb) Updatedc) Deleted
How these events are sequenced
How to define a sequence?
What events exactly?
HOW CAN WE FIND THEM?
How to DEFINE interesting activities?
SEQUENTIAL AND TEMPORAL ANALYSIS(e.g. Johnsson, D’Mello & Azevedo, 2012; Clearly, Callan & Zimmerman, 2012;
Molenaar & Järvelä, 2013; Malmberg, Järvenoja & Järvelä, 2013)
on – line chat
discussions
log file traces
Learning task
outcomes
1234
SSRL (f=35)SRL (f=321)
1. BEFORE2. DURING3. AFTER
SRL SRLSSR
L
SRL SRL SRL SRL
MICROLEVEL EXAMPLES
1
2
3
4
5
UNIT OF ANALYSIS: TIMING OF SRL + SSRL
EVENTS
DATA ANALYSIS – LOG FILE TRACES… (nStudy, Winne et al., 2010)
Trace data activity Theoretical definition
Frequency
Internal actions 248
View Task Instruction (TI) Task understanding 101
View Planning Note (VP) Task understanding 60
View Edited Planning Note (VEP)
Monitoring 56
View Edited Reflection Note (VER)
Monitoring 31
Interactive actions 73
Edit Reflection Note (ER) Evaluating 45
Edit Planning Note (EP) Planning 28
DATA ANALYSIS – ON-LINE CHAT DISCUSSIONS
Coded SSRL episodes (6188 lines)(meaningful collective interaction chat episodes; Greeno, 2006)
Socially shared task understanding 3
Socially shared planning 17
Socially shared strategy use 4
Socially shared motivation 11
Total SSRL codings 35
LEARNING TASK OUTCOMES
TASK OUTCOME M
1 4 3 2 1 1 2.2
2 2 1 1 3 4 2.2
3 2 3 1 3 4 2.6
• Collaborative learning outcomes from each three learning task among the five groups were coded and categorized on a likert scale varying from 1 to 4 (Biggs 1984).
1= lowest, 4= highest score
MICROLEVEL DATA EXAMPLEIntegration of coded chat and log data
INTERNAL INTERACTIVE MICROLEVEL SEQUENCE OF SHARED REGULATION
Task Understanding Socially shared strategy
Task understanding
Socially shared
strategy
VP TI TI TI SSTR TU SSTR
….This is how it looks in analytical level…
Self-regulated learning:TI=Task InstructionsVP= View Planning
Socially Shared Regulation:SSTR= Socially shared strategy
+ =
RESULTS 3What characterizes temporal sequences of self- and shared
regulation activities in high – and low learning outcome situations?
• Example 1 of self- and shared regulation in HIGH learning outcome task
• Example 2 of self- and shared regulation in LOW learning outcome task
BEFORE DURING AFTER
1 TU SSTU SSPL 1 TU PL 1 SSM REF 2 MON TU PL 2 MON REF SSPL 3 TU SSM 3 MON REF 4 MON
BEFORE DURING AFTER
1 TU PL 1 MON TU MON 1 REF 2 MON 2 TU MON
TU=Task understanding: MON= Monitoring; PL= Planning; REF= Reflection; SSPL= Socially shared planning; SSM=Socially shared motivation regulation
TO CONCLUDE…
• Analysing temporal sequences of SRL and SSRL informs regulatory processes as they unfold
• Simplified patterns inform about changes in regulation in contrasting cases
• Important details can be lost?• Generalisation of findings to different
settings?
OTHER ANALYTICAL APPROACHES
• State lag sequential analysis (e.g. Bakeman & Quera, 2007)
• Process discovery (Gunther & VandeAalst, 2007)– Fuzzy mining algorithm
• Data mining & parsing (e.g. computer generated logfile traces) (Romero & Ventura, 2007)