Upload
duongxuyen
View
219
Download
0
Embed Size (px)
Citation preview
PeerWise Monash, December 2007 – 1
PeerWise and Contributing StudentPedagogy
John Hamerwith Paul Denny and Andrew Luxton-Reilly
Department of Computer ScienceUniversity of Auckland
Talk given at Monash University, 10 December 2007
Introduction
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 2
Contributing Student Pedagogy
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 3
■ Students create and share learning resources; e.g.notes, visualisations, instructional videos, quizquestions, reading lists
■ Web-based collaboration tool (e.g. wiki) used tostore work-in-progress and share course material
■ Peer feedback and evaluation
■ Related theories: flexible learning (Collis),constructivism, community of practice (Wenger),ZPD (Vygotsky)
What is PeerWise?
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 4
■ An online bank of multiple choice questions
■ All content is student generated
Features
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 5
Not just questions and answers
■ explanations
Features
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 5
Not just questions and answers
■ explanations
■ responses
Features
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 5
Not just questions and answers
■ explanations
■ responses
■ discussion threads
Features
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 5
Not just questions and answers
■ explanations
■ responses
■ discussion threads
■ difficulty and quality ratings
Features
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 5
Not just questions and answers
■ explanations
■ responses
■ discussion threads
■ difficulty and quality ratings
■ leader-boards
Learning
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 6
■ Designing a question
◆ focuses attention on learning outcomes
◆ encourages reflection on course material
■ Choosing distractors
◆ misconceptions are considered
◆ promotes deep understanding
■ Writing explanations
◆ students express understanding in their ownwords
Learning
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 6
■ Answering questions
◆ useful practice / revision
◆ reinforces learning
■ Evaluating quality
◆ requires critical analysis
■ Providing feedback
◆ encourages peer dialogue around learning
Learning
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 6
Probably setting up my multi-choice question. This
was pretty hard given that i had to think of the
possible wrong solutions students would fall for and
required a lot of thinking from me, which in the end
was a lot of help because i was just about able to
answer any question that was on the same topic as
my question.
That was the biggest learning experience for me!
— OE4/205
For instructors
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 7
■ Large question bank developed at low cost
◆ ENGGEN 131 (introductory programmingcourse)
◆ 570 students
◆ 1,700 questions
◆ 35,000 responses
◆ Sept 9th – Nov 1st
■ Can be used as a basis for exam questions
For instructors
IntroductionContributing
Student Pedagogy
What is PeerWise?
Features
Learning
For instructors
Results
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 7
■ assists staff inidentifyingweaknesses
■ reveals how wellstudents areengaging withcertain topics
Results
Introduction
Results
Courses
Overall use
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 8
Courses
Introduction
Results
Courses
Overall use
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 9
■ Five courses used PeerWise in 2007
■ Varied in: coursework marks awarded for theactivity (Worth), questions to write and answer,use of MCQs in exam and tests, and whether thePeerWise authors taught the course.
Course Worth Write Answer Exam PW taught?
CS101 2% 2 10 Both YesCS111 1% 2 2 None YesCS105 2% 2 10 None NoCS220 1% 2 10 Both NoEG131 7% 2 20 Exam Yes
Overall use
Introduction
Results
Courses
Overall use
ENGGEN 131
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 10
CourseMean
QuestionsMean
AnswersMean
Comments
CS101 2.9/2 37.5/10 13.5CS111 1.9/2 60.7/2 5.0CS105 2.4/2 47.0/10 9.2CS220 1.6/2 36.0/10 7.2EG131 3.0/2 62.6/20 19.1
ENGGEN 131
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 11
Course details
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 12
■ Compulsory for all first-year engineering students
■ Two independent 6-week sections: MATLAB, C
■ PeerWise introduced only in the C section
■ MCQs in C section of the exam
■ Collected individual data on: number ofquestions written (Qs); questions answered (As);length of comments (NC); average commentlength (AvgC); MCQ (i.e. C programming) mark;MATLAB mark
Grade distributions for MATLAB and C
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 13
0 100 200 300 400 500
2040
6080
Gra
de
Grades distributions for MATLAB (red) and C (blue)
C vrs. MATLAB
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 14
0 20 40 60 80 100
020
4060
8010
0
MCQ grade
MA
TLA
B g
rade
Grades in the two sections are similaridentical mean p = 0.11
Linear regression model for participation
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 15
0 20 40 60 80 100
020
4060
8010
0
MCQ exam grade
Pre
dict
ed M
CQ
= 0
.12A
s +
12Q
s −
0.1
4NC
+ 0
.33A
vgC
Predicting MCQ grades from As+Qs+NC+AvgC
Observations
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 16
■ The linear regression model came up with theprediction
MCQ = 0.125As+12.05Qs−0.14NC+0.33AvgC
■ The largest contribution comes from Qs, thenAvgC, followed by As and (negligibly) NC. Theratios for an average student are: 31 : 16 : 8 : −3
■ Writing questions is the single most significantpredictor of exam performance, followed by theaverage length of comments.
■ These are both activities associated with deeplearning
C-MATLAB gain by quartile
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 17
Q1 (low) Q2 Q3 Q4 (high)
−20
020
Diff
eren
ce M
CQ
−M
AT
LAB
Gain in MCQ−MATLAB by participation quartile
High/low participation C-MATLAB gain
Introduction
Results
ENGGEN 131
Course detailsGrade distributions
for MATLAB and C
C vrs. MATLAB
Model
Observations
Gain by quartile
Gain by median
PCA
Gender data
Questions?
PeerWise Monash, December 2007 – 18
Below median Above median
−20
020
Diff
eren
ce M
CQ
−M
AT
LAB
PeerWise participation correlates with gain95% interval = 1.6−5.5, p = 0.0004
PCA
Introduction
Results
ENGGEN 131
PCA
Factor map, all data
Individuals, all data
Individuals,
minimum Qs
Gender data
Questions?
PeerWise Monash, December 2007 – 19
Factor map, all data
Introduction
Results
ENGGEN 131
PCA
Factor map, all data
Individuals, all data
Individuals,
minimum Qs
Gender data
Questions?
PeerWise Monash, December 2007 – 20
−1.0 −0.5 0.0 0.5 1.0
−1.
0−
0.5
0.0
0.5
1.0
Variables factor map (PCA)
Dimension 1 (36.29%)
Dim
ensi
on 2
(24
.56%
)
As
Qs
NC
AvgCMCQExam
Individuals, all data
Introduction
Results
ENGGEN 131
PCA
Factor map, all data
Individuals, all data
Individuals,
minimum Qs
Gender data
Questions?
PeerWise Monash, December 2007 – 21
−5 0 5 10 15
−6
−4
−2
02
4Individuals factor map (PCA)
Dimension 1 (36.29%)
Dim
ensi
on 2
(24
.56%
)
1
234
56
7
8910
11
12
13
14
151617
1819202122
2324252627
28
29303132333435
36
37
38
3940
41424344
4546474849
50
52
53
54
55
5657
58
59606162
63
64
65
66
6768
697071
727374
75
7677
78
79
80
81828384
85
86 8788
8990
91
9293
94
95
96979899100101 104
105106
107108
109110111112
113114
115116117
118
119120
121
122
123124
125126
127128
129
130
131
132
133134135
136
137
138139140141
142143
144145146147148
149150151152
153
154
155
156
157158159160
161162163164
165166167
168169170
171172173
174
175176177
178
179
180
181182
183184
185
186187188189
190191192
193194195
196197
198199
200
201
202
203
204
205
206
207
208
209210
211
212
213
214
215216217218
219
220221222
223224
225226
227
228
229
230
231
232233
234
235
236
237238
239
240
241
242243244
245
246
247248249250251252253254
256
257
258
259
260261
262
263
264265
266
267268269
270
271272
273274
275
276
277278
279280281282
283
284
285
286
287
288
289
290
291
292
293
294
295296297298299
300301302303304
305
306
307
309
310311
312313314
315
316
317
318
319
320
321322
323324
325326
327
328
329
330
331332333334
335336
337
338
339
340
341342
343344
345
346347348
349
350351352353
354
355
356
357358
359360361
362
363364365
366 367
368369370371
372373374375
376377
378
379
380381382383
384385386
387
388
389
390
391392
393394 395
396
397
398399400401402403404
405
406407
408
409
410411
412413
414415416417
418
419420
421422423
424425426427428
429430431432
433
434
435
436437
438
439
440
441442443444
445
446447
448
449450
451452453
454455456
457
458
459
460461
462
463464
465
466467
468469
470
471472
473474
475
476477
478
479
480
481
482
483484
485486487
488
489
490
491
492
493494495496
497498
499500501
502
503
504
505506507
508509
510511512
513514
515516517518
519
520521
522523
524525
526
527
528529530
531
532
533
534535536537
538539540541542543544
545546547548549
550551552553554
555556557
558559
Individuals, minimum Qs
Introduction
Results
ENGGEN 131
PCA
Factor map, all data
Individuals, all data
Individuals,
minimum Qs
Gender data
Questions?
PeerWise Monash, December 2007 – 22
−5 0 5 10
−6
−4
−2
02
4Individuals factor map (PCA)
Dimension 1 (37.19%)
Dim
ensi
on 2
(25
.2%
)
1
2
4
6
1013
1617
2223
2427
28
2930313334
36
37
39
42434445
4647484950
53
5760
63
64
65
6768
697071
7478
80
828384
85
8688
8990
91
93
94
95
979899100101
106
107108
118
119
123124
126
127129
130
132
133
136
137
139140141
142143145147148
149150
151 155156
161162163
169170
173
174
176177
180185
187
189
190191192
193
194195
196197
199 200
201
202
203 207
208
209210
213
214
219
220221
222224
225227
228
229
230
231
232
234
236
237238
240
241
242243
245
246
247249250251253254
255
257
263
265266
267269271
274276
281
282
283285287
288
289
290
291
292
293295297298299
300301302303304
305
308309
310312313314316
317
318
321322
323
325328
329
330
331332334
335
337
338
339
340
341343344
345347
349
350351352353
354
355
356 359361
362
368
370371376
378
379
381385386
387
388390
391392
393394 395
396
397
399400401402403406
408410
412
415416417
418
419420
421422423
424425426427
429430431432
433
434
435437
438
440
441442443 444
445
447449450
451452453
454455
457
458
461
462
463465
466467
468469
470472
473474
475
476477
478
479
480
481
482
483484
486487
488
489
490491
493495496
497498
499500
501
502
503
504507
508
511
513516518
519
521522523
524525
527528530
531
534535537538540
542543 545546547548
549 550551
552553
554
555556557558
559
Gender data
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 23
Males slightly outperform females
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 24
Female Male
2040
6080
MC
Q e
xam
mar
k
Men outperformed women in C95% interval = 2.7−9.4, p = 0.0005
Males slightly outperform females
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 24
Female Male
3040
5060
7080
90
MA
TLA
B e
xam
mar
k
Men do better in MATLAB, but not to the same extent95% interval = 1.3−6.2, p = 0.003
No gender difference in PeerWiseparticipation
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 25
Female Male
02
46
810
1214
Que
stio
ns p
oste
d
Men and women both contributed Qs equallyp = 0.7
No gender difference in PeerWiseparticipation
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 25
Female Male
050
100
150
200
Que
stio
ns a
nsw
ered
Men and women both answered Qs equallyp = 0.6
No gender difference in PeerWiseparticipation
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 25
Female Male
020
4060
8010
0
Tot
al c
omm
ents
(ch
ars)
Men and women both posted comments equallyp = 0.27
No gender difference in PeerWiseparticipation
Introduction
Results
ENGGEN 131
PCA
Gender dataMales slightly
outperform females
No gender difference
in PeerWise
participation
Questions?
PeerWise Monash, December 2007 – 25
Female Male
050
100
150
200
250
Ave
rage
com
men
t len
gth
Men and women both wrote similar length commentsp = 0.13