Honing in on Social Learning in MOOC Forums: Examining Critical Network Definition Decisions Alyssa Wise, Yi Cui, Wan Qi Jin LAK’17, March 2017, Vancouver, CANADA
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Honing in on Social Learning in MOOC Forums: Examining Critical Network Definition Decisions
Alyssa Wise, Yi Cui, Wan Qi JinLAK’17, March 2017, Vancouver, CANADA
Investigation of the interaction practices in large-scale learning environments based on analysis of the artifacts left behind by learners’ and instructors’ activity.
MOOCEOLOGY
2
QUESTIONS DATA METHODS RESULTS CONCLUSION
RQ1: What effects do different tie definitions have on network characteristics?
RQ2: In what ways do unpartitioned, content-related, and non-content social networks show distinct characteristics?
RQ3: What differences in the discussion interactions may account for the distinctions between networks?
Research Questions
2
Learning Context & Data • A MOOC on statistics in medicine
• 3-level posting structure
• Forums facilitated by 2 instructional team members
• 567 users
• 817 threads containing 3124 posts 4
QUESTIONS DATA METHODS RESULTS CONCLUSION
Content-based Partitioning• Content-related = Q&As/comments related to the course subject • Non-Content = Others (logistical & technical Q&As, socilizing) • Method: Dynamic Interrelated Post and Thread Categorization
(Cui, Jin, & Wise, in press), est. accuracy = .88
5
Non-content
Content-re-lated
Threads
468 threads
349 threadsContent-related activities
17831%
Both15728%
Non-content activities 23241%
PARTICIPANTS
QUESTIONS DATA METHODS RESULTS CONCLUSION
6
S1 R1
R2
RR1
RR2
RR3
S1 R1
R2
RR1
RR2
RR3
S1 R1
R2
RR1
RR2
RR3
S1 R1
R2
RR1
RR2
RR3
S1 R1
R2
RR1
RR2
RR3
Direct Reply Star Direct Reply + Star
Limited Copresence
Total Copresence
RQ1: Tie Definition Effects
QUESTIONS DATA METHODS RESULTS CONCLUSION
7
Unp
artiti
oned
Cont
ent-r
elat
edN
on-c
onte
nt
Number of edges Avg. node degree Avg. edge weight
0
1000
2000
3000
4000
5000
6000
0
5
10
15
20
25
0
5
10
15
20
25
Direct Reply Star Direct Reply + Star Limited Copresence Total Copresence
0
1000
2000
3000
4000
5000
6000
0
5
10
15
20
25
0
5
10
15
20
25
0
1000
2000
3000
4000
5000
6000
0
5
10
15
20
25
0
5
10
15
20
25
Tie Definition Effects - Resulting Network Properties
Direct Reply Star Direct Reply + Star Limited Copresence Total Copresence
Unp
artiti
oned
Cont
ent-r
elat
edN
on-c
onte
ntTie Definition Effects - Resulting Networks
Direct Reply Total CopresenceU
npar
tition
edCo
nten
t-rel
ated
Non
-con
tent
0
5
10
15
20
25
0
5
10
15
20
25
0
1000
2000
3000
4000
5000
6000
0
1000
2000
3000
4000
5000
6000
0
5
10
15
20
25
0
5
10
15
20
25
0
1000
2000
3000
4000
5000
6000
0
5
10
15
20
25
0
5
10
15
20
25
Number of edges Avg. node degree Avg. edge weight
Direct Reply Total Copresence
0
1
2
3
4
5
6
7
8
Avg edge weight0
1
2
3
4
5
6
7
8
Avg node degree
10
Content-related network(# of nodes = 335, # of edges = 848)
Non-content network(# of nodes = 389, # of edges = 724)
RQ2 & 3: Content vs Non-Content Networks
Content-related Non-content
QUESTIONS DATA METHODS RESULTS CONCLUSION
CL1 NL1 NL20123456789
10
Avg node degree
CL1 NL1 NL202468
10121416
Avg edge weight
11
Learner Modules
CL = Content-related Learner Module NL = Non-content learner module
CL1(# of nodes = 23, # of edges =57)
NL1(# of nodes = 62, # of edges =71)
NL1(# of nodes = 23, # of edges = 28)
QUESTIONS DATA METHODS RESULTS CONCLUSION
12
Content vs Non-Content Discussions1. Thread Characteristics
Content-related Module 1
Non-content Module 1
Non-content Module 2
# of contributing threads 30 2 11
Avg # of posts per participant in thread 2.41 (1.62) 1.13 (0.07) 1.42 (0.53)
Avg # of posts per thread (SD) 13.4 (14.85) 6, 93 6 (5.75)
2. Activity characteristics: content-related activities involved more• complicated topics• interaction techniques • social presence cues
QUESTIONS DATA METHODS RESULTS CONCLUSION
13
U225: Congrats [u10]! Yes, it has been hard, but fun, and we learned an awful lot, right?U110: Great! Everyone it was a pleasure to work with you. Thank you….U10: YES [u225]! And [u110] - the test was scary - I thought of my discussion board friends often!!U216: Thanks, thanks so much to [u10], [u152], [u110], [u225] and everybody who helped us to understand this beautiful course! And in my case also for writing many posts, I see I have improved my English skills and my statistics vocabulary!!!U225: [u10], [u216], [u152], [u110], [u515] and everyone, your discussions helped me so much. I was always a few days behind you in homework - glad I was able to catch up in the last weeks and participate a little bit….
Content-related learner module 1
Examining Learner Interaction
U225: [u10], [u216], [u152], [u110], [u515] and everyone, your discussions helped me so much. I was always a few days behind you in homework - glad I was able to catch up in the last weeks and participate a little bit….
U10: YES [U225]! And [u110] - the test was scary - I thought of my discussion board friends often!!
QUESTIONS DATA METHODS RESULTS CONCLUSION
14
CI1 CI2 NI1 NI2
# of nodes (% in network) 184 (54.93%) 75 (22.39%) 168 (43.19%) 47 (12.08%)
# of edges (% in network) 400 (47.17%) 105 (12.38%) 315 (43.51%) 55 (7.6%)
Avg node degree (SD) 4.35 (11.06) 2.8 (7.56) 3.75 (11.18) 2.34 (6.03)
Avg edge weight (SD) 2.23 (3.21) 1.83 (1.72) 2.11 (2.48) 1.20 (0.44)
Instructor Modules
CI = Content-related instructor module NI = Non-content instructor module
CI1 CI2 NI1 NI2
QUESTIONS DATA METHODS RESULTS CONCLUSION
U1 • Responses at all levels• Coaching and supporting • Social presence cues
U417• Responses to thread starters• Straight forward answers • Little social presence
15
Comparing Intervention Approaches
QUESTIONS DATA METHODS RESULTS CONCLUSION
“Think about it again using the hint and let me know if you have any other questions.”
“That is correct - Nice! So how would you use this to solve the question?”
“A bell shape is not necessary. You could have a 'bimodal' distribution where the two groups do not follow a bell shape.”
Conclusions• Tie definition affects network and interpretation.
• Content-related and non-content networks had distinct characteristics.
• Content-related discussions led to wider and deeper learner connection; instructor’s intervention approach may also affect it.
16
QUESTIONS DATA METHODS RESULTS CONCLUSION
Related Documents
