what they did really do with the social software and mobile devices? a temporal analysis of...
TRANSCRIPT
"Exploring methodological approaches to conduct process analysis
in fragmented, long-duration collaboration mediated by multiple
artifacts."
Jari LaruLearning & Educational Technology Research Unit
University of Oulu10/2009
Pedagogical design..
Syndication service via RSS
Blog
Theoretical grounding
Groups’ grounding
Reflection,Constructing
Reflection,Constructing
Reflection,Constructing
Reflection,Constructing
Elaborating Constructing
Media-sharing services
Course wiki & Blog (announcements; schedules; lecture materials; workspaces; helpdesk)
Wiki
Media
Media
Media
Media
= Face-to-Face sessions
Meaning-making sessionCollaborative virtual work
Individual virtual work
Grounding sessionLecture
Monitoring
Theoretical grounding
led to messy data..
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Reflection,Constructing
Reflection,Constructing
Reflection,Constructing
Elaborating
time
Constructing
video recordings
media artefacts
media artefacts
video recordings
Blog content Wiki historylecturetopics
pre-test (conceptual knowledge)
+ questionnaire
post test SRI interviews
video recordings
RSS Feeds RSS Feeds
monitored content
conversationanalysis
Conversationanalysis
Content analysis
statisticalmethods
Conversationanalysis
Content analysis
Content analysis
Content analysis
Content analysis
statisticalmethods
statistical methods used to calculate dependencies or probabilities between variables
Variable level
Event level
Ouput: a lot of different kind of frequency data a la coding categories, log data, conceptual test etc
Research problem was and is uh..challenging
What they did really do with the social software and mobile devices? A temporal analysis of long-term technology mediated collaboration.
Methodological skills oriented my interests towards variable
level approach
• Content Analysis
• Social Network Analysis
• Conversation Analysis
• Mann-Whitney U-test
• etc..
Variable level
Well, what’s the problem with variable data?
We werent’ only using MULTIPLE tools..
Blog Wiki
but we also had different kind of F2F settings in different phases
Variable level
Actually results lived in data archipelago, were small (or
big)islands there
Having a holistic approach into analysis and design, it was little
bit too tricky to start..
Ishikawa diagrams(also called fishbone diagrams or cause-and-effect diagrams)
First event-level try-out. No success. No methodological bacground. FAIL
Event level
FAIL
A Framework for Analyzing Interactional Processes in Online Learning
Daniel Suthers, Nathan Dwyer, Richard Medina, Ravi Vatrapu (2007). Paper presented at the Annual Meeting of the American Educational Research Association (AERA), Chicago, April
9-13, 2007.
FAIL
Too instrumental & multi-widget-chat tool orientation (several presentations at CSCL’09)
Event level
What’s next?• Didn’t found appropriate methodological
literature of event level analysis
• wrong terminology used (didn’t know)
Event level
WEB2.0 / Social WEB• this terminology didn’t help neither
• Methodologically exteremely heterogenous area
Variable level approachwas chosen..
Appropriate coding categories?
FAIL OK
Blog Wiki
Cress & Kimmerle (2008) A systemic and cognitive view on collaborative knowledge building with wikis. International Journal of Computer-Supported Collaborative Learning, 3(2), 105–122.
OK
Basic categories:“Besides assimilating information the wiki can also accommodate. This happens when new information is not only attached to the existing information, but the information in the wiki is organized in a new way.”
Problem remains
Data
Scheme
Variables
Data
Scheme
Variables
Data
Scheme
Variables
Blog Wiki OtherFAIL FAIL
Analytical toolkit for time-series data
MISSINGWiki: timetags are essential. No way to import into
NVIVO8Blogs: cumulative nature and “metadata” essential. No
way to import into NVIVO8
WHERE IS A TOOL FOR ANALYSING HETEROGENOUS, MULTITOOL TIME-SERIES
DATA? Answer: Tatiana! (CSCL 09)
TIME IS RUNNING!New tryout with variable
level approach
•How to measure dependencies between (all) variables??!
•Petri Nokelainen visited our group: IDEA! Bling!
Bayesian dependency modeling• SRL Self-report questionnaire (Pintrich, 1993; Wolters, 2001)
• Short-answer (paper & pencil) conceptual knowledge test
– 3 x 6 concepts (about lectures), same in pre/post tests
• Basic activity measures as “log data”
– # of images/videos (file-sharing services)
– # of articles/edits (blog, wiki)
– # of wordcount (blog, wiki, face-to-face sessions)
– # of items (rss)
• Qualitative data collection for other analyses (stimulated recall interviews, group interviews, content produced by learners, wiki history)
• Statistical dependencies between the variables were studied with Bayesian Dependency Modeling (BDM), which predicts the most probable statistical dependency structure between the observed variables (Myllymäki, Silander, Tirri & Uronen, 2002).
• B-Course: A Web-Based Tool for Bayesian and Causal Data Analysis was used as a analysis toolkit
• Bayesian analysis is based on propabilities instead of frequencies (Nokelainen, Silander, Ruohotie, Tirri, 2007)
var 1 var 2B-Course
LET - Learning and Educational Technology Research UnitJari Laru. 27.8.09 ERLI09, Amsterdam
Are learners’ actions in the web2.0 tools and F2F sessions related to their learning outcomes?
19
“monitor”
“reflector”
Some arcs are pretty strong, so that removing any of them would result in a model with probability less:* than one millionth of that of the original model** than one thousandth of that of the original model*** smaller than one thousandth (weak connection)
blog
mediasharing
Syndication service
wiki
wiki
direct dependencies:Learning gain:- RSS_Read (# of read RSS items)=> Monitor- WP_Entries (# of blog entries)=> Reflector
Learning outcome
FAIL
LET - Learning and Educational Technology Research UnitJari Laru. 27.8.09 ERLI09, Amsterdam
WHY IT WAS DOOMED TO BE A FAIL?
• Bayesian approach provided only surface level analysis:–deep insight into collaborative interaction in F2F groups, wikis & blogs missing
–order, duration and quality of events: eg. lecture > F2F > image/video > blog reflection > F2F > Wiki
• Interesting question “How learning unfolds over time” wasn’t answered at all..
• Didn’t support design of the study!
20
OOPS!
What’s next?
What I had on my hands after Bayesian?• Analytical toolkit: Tatiana
• Some coding schemes which didn’t stretch over the whole design
• Some methodological hints about event based approach..
• Intuition that event based approach is “my way”
Narrative and learning with web2.0 technologies: towards a research agenda.
Pachler & Daly (2009). Journal of Computer Assisted Learning. 25, 6-18
•Narrative trails (+)•Sense-making (+)•Blog centric (-)•vague understanding of collaboration (-)(commenting in blogs)•Holistic approach?..
Well.. hmm
Task level
Missing piece is ..
Missing piece
Time is precious: Variable- and event-centred approaches to process analysis in CSCL research.
(2009) Computer-Supported Collaborative Learning (2009). 4:239-257
Methodological link
Suther et. al (2007)graphical techniques
Mercer (2008) et. alcommunication &
learning in classroom
Muukkonen et. al (2008)
Time series analysis
Central tenet: “methods developed in conversation analysis and ethnomethodology...do not carry over to interactions that are
fragmented over time, stretch over longer durations, or are mediated by artifacts rather than talk” pp.241
Process method should.. (Reimann, 2009)
• a process method should, in addition to efficient cause (typical for event approach), be able to deal with at least two other kinds of causes:
• formal cause, referring to the patterns of which things are made
• final cause, the end for which things are made [goal, in our study: wiki]
One coding scheme to evaluate interpersonal exchange
• Mercer, N. (2008). The seeds of Time: Why Classroom Dialogue Needs a Temporal Analysis. The Journal of Learning Sciences, 17: 33-59
• Focus of article is on examining long-term classroom talk and how it’s used to:
• represent past experiences
• carry ideas forward from one occasion to another
• approach future activities
• achieve learning outcomes
I propose an applied approach for Mercer
• To evaluate interactions that are mediated by artifacts or than talk and can be categorized according Mercer:
• represent past experiences [for example: in personal blog, or F2F discussions]
• carry ideas forward from one occasion to another [photo => blog, or peer blog => own blog etc.]
• approach future activities [design groups’ / individual activities]
• achieve learning outcomes [troughout the design]
Unit of analysis in my research
• Main entities are individuals and groups
• events are constrained to those incidents in which either individuals or groups can participate
• Focus will be put on the sequences of activities, incidents, crises, or stages that unfold in the groups over the duration of the course
led to messy data..
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Theoretical grounding
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Media
Reflection,Constructing
Reflection,Constructing
Reflection,Constructing
Elaborating
time
Constructing
video recordings
media artefacts
media artefacts
video recordings
Blog content Wiki historylecturetopics
pre-test (conceptual knowledge)
+ questionnaire
post test SRI interviews
video recordings
RSS Feeds RSS Feeds
monitored content
conversationanalysis
Conversationanalysis
Content analysis
statisticalmethods
Conversationanalysis
Content analysis
Content analysis
Content analysis
Content analysis
statisticalmethods
statistical methods used to calculate dependencies or probabilities between variables
Variable level
Event level
Ouput: a lot of different kind of frequency data a la coding categories, log data, conceptual test etc