Upload
jaidyn-helin
View
213
Download
0
Tags:
Embed Size (px)
Citation preview
UEM College 9 mei, 2006
The Paradox of the Assisted User: Guidance can be Counterproductive
Christof van Nimwegen
UEM College 9 mei, 2006
Having a plan…..the basis of success?
O’Hara & Payne: Plan-based : Use strategies from long term memory Display based: Rely on interface information
Making an interface more costly provoked - More plan-based strategies- Better transfer to same domain problems
Zhang 1997:Externalizing information can be useful for cognitive tasks; the more information is externalized, the easier it is to
solve the problem
UEM College 9 mei, 2006
Background: Externalization vs. Internalization
Important for interaction designhow much information should be available on the interface (externalization) as opposed to having users encode and store this information in memory (internalization).
We investigate conditions whereby externalizing the information in the interface, or requiring it to be internalized respectively, enhances task performance of users.
This research contributes to formulate guidelines for GUI design in the context of problem solving tasks.
UEM College 9 mei, 2006
Externalization vs. Internalization: Example
Externalization Representation with relevant feedback Remembering certain task-related
knowledge is unnecessary This relieves working memory
(low cognitive load)
Internalization No support: Task-related information
is unavailable on the interface It needs to be internalized (high
cognitive load) and stored in memory Knowledge in memory of user is more
solid and detailed (recall, not recognition)
UEM College 9 mei, 2006
General research questions
In literature, there is no real agreement, but externalization in modern GUI’s is more or less default
But does this “support” of externalization … really result in better “performance” after all? make a task more pleasant in the long run? help when “learning something” is the aim?
(think of software in education) with transfer tasks? result in better knowledge of the underlying
task? what about interruptions?
UEM College 9 mei, 2006
Earlier experiment: Problem choiceClassical: Missionaries & Cannibals
Bring all of them to the other side
Rules: - Maximum 3 in boat - Minimum to sail is 1 (not empty)- Cannibals never in the majority or they
will eat the missionaries
Abstract version: Ball & Boxes
Transfer: Realistic M&C
UEM College 9 mei, 2006
Path
UEM College 9 mei, 2006
Material first experiment (with greyed-out items)
Abstract puzzle: variation on Missionaries & Cannibals
INTEXT
A more realistic transfer task: Missionaries & Cannibals
UEM College 9 mei, 2006
Results experiment 1
Internalization resulted in better problem knowledge afterwards better performance after long interval better performance on transfer task after
interval
Externalization had no advantages at all!
UEM College 9 mei, 2006
Externalization and planning?
Externalization
> Relying on interface discourages planning > Shallow thinking
> Worse performance
Internalization
> Stimulates planning> More engagement
> Better performance
UEM College 9 mei, 2006
Second experiment: more realistic task
Independent variables
Interface style (INT-EXT)We used more realistic material. Instead of abstract puzzles a conference planning task
Cognitive styleWe included an attitude trait. Before we tried to externally induce some state of mind, now we want to look at stable trait in people. We chose for the “Need for Cognition” (NFC). NFC is a well proven construct measuring the tendency of individuals to engage in and enjoy effortful cognitive tasks
UEM College 9 mei, 2006
Hypotheses
H1: Internalization leads to a more plan-based strategy and better performance than externalization. Having information externalized tempts users not to form plans and rely on the interface. The internalized condition lacks this guidance and is encouraged to plan and think before acting.
H2: High NFC leads to a more plan-based strategy than low NFC. People with high NFC have high intrinsic motivation to think and engage in effortful cognitive tasks, will show more plan-based behavior, and perform better
UEM College 9 mei, 2006
Method
Subjects 43 subjects, 17 (40%) male, 26 (60%) female. Aged 19 to 32 years Following or recently having followed higher education The experiment took at most one hour, 5 euro reward
Design 2*2Independent variables: Interface style (INT <> EXT)
Cognitive style (HIGH <> LOW) (median split)
Dependent variables: Various performance measures (times, moves)
StrategyKnowledge & Opinions
UEM College 9 mei, 2006
The application
The “Conference Planner” application simulated the planning of speakers (it logged all the moves)
Each of the speakers had different demands. They had to be scheduled in one of three conference rooms
The conference rooms each had its own facilities and
availability
Without planning, the scheduling will not be optimal and extra moves besides the optimal path must be made
The difference between EXT and INT was implemented by providing information concerning the legality of moves/actions
UEM College 9 mei, 2006
Conference planner: externalization and internalization
In EXTERNALIZATION it showed where speakers can be placed.
Slots in the timetable (satisfying constraints being available) turn green
Not the best slot, just legal ones
In INTERNALIZATION the interface, this feedback was absent
Subjects need to look more atthe constraints and think and plan themselves
UEM College 9 mei, 2006
Measures…..
UEM College 9 mei, 2006
Procedure
1. NFC2. Conference Planner
a) Trial 1b) Trial 2c) Trial 3d) Trial 4e) Trial 5
3. Post Experimental Questionnaire
UEM College 9 mei, 2006
Results: Time based measures (av. per 5 trails)
Time before first moveINT took longer than EXT
(p<.05)INT:M=19.8, SD=7.4EXT:M=15.3, SD=7.2
High NFCLow NFC
Tim
e b
efo
re f
irst
mo
ve
24
22
20
18
16
14
Internalisation
ExternalisationHigh NFCLow NFC
T
ime
be
twe
en
mo
ves
5.5
5.0
4.5
4.0
3.5
Internalisation
Externalisation
Time between movesINT took longer than EXT
(p<.05)
INT: M=4.8, SD=1.4EXT: M=3.9, SD=1.3
Total time: no significant difference
UEM College 9 mei, 2006
Results: Move based measures (av. per 5 trails)
Superfluous moves:INT made fewer than EXT
(p<0.05)INT: M=2.5, SD=0.6EXT: M= 4.3, SD=0.6
High NFCLow NFC
S
up
erf
luo
us
Mo
ves
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
Externalisation
Internalisation
Reconsidered moves: INT made fewer than EXT
(p<0.01)INT: M=0.5, SD=0.4EXT: M=2.1, SD=0.4
High NFCLow NFC
R
econ
side
red
mov
es
3.0
2.5
2.0
1.5
1.0
.5
0.0
Externalisation
Internalisation
Invalid moves : no significant difference
UEM College 9 mei, 2006
Strategy & opinion
Internalization subjects used the ‘most constraints first’ strategy more often. Tendency, (F(1,39)=3.21, p=0.08).
INT : 2.4 times out of 5 (SD=1.76) EXT : 1.5 times out of 5 (SD=1.5)
“I sometimes did not know how to proceed with the arrangement of the speakers”, (F(1,39)=5.91, p<0.02).
EXT :M= 5.24, SD =1.61INT :M= 6.23, SD=0.97)
(the item was scored reversely)
UEM College 9 mei, 2006
Conclusion
Hypothesis accepted: The interface requiring more internalization
resulted in
longer thinking time before subjects started more time between moves fewer superfluous moves and deviations, thus with
greater efficiency
This indicates that when information has to be internalized, more planning, active thinking and contemplation from the users is provoked
UEM College 9 mei, 2006
Conclusions
H2: rejected. Cognitive style along the dimensions of NFC had no influence.
It was Interface style again that predicted problem solving behavior
quite strongly, and could overrule possible effects of pre-existing individual differences. Also, our subjects were volunteering
university students, where the NFC average was very high.
UEM College 9 mei, 2006
Conclusions
If plan-based behavior is preferred, designers should be careful in giving users too much assistance (what is the goal?)
With multimedia being present in all corners of society, our findings can be valuable in the development of applications in the realm of education, multimedia learning or game based learning
UEM College 9 mei, 2006
Further research
Understanding reactions to interface information (based on cognitive findings) is important in tuning software, thereby allowing it to achieve its goal.
We will continue the research, and broaden the types of problem solving activities, and include other variables related to attitudes to problem solving.
UEM College 9 mei, 2006
Future research
Severe Interruptions(PC Crash)
Transfer Task(same task, different feeling)
Use eye tracking(distraction? Where are they really
looking?)
UEM College 9 mei, 2006
Deel 2Wat jullie gaan doen!?
Deel 2
UEM College 9 mei, 2006
Week 20
Maandag 13-15 Werkgroep Usability lab (CGN,C201)
Groep 1: 13:00Groep 2: 13:30 4 proefpersonenGroep 3: 14:00 4 proefpersonenGroep 4: 14:30 6 proefpersonen
_____________ 14 pp totaal
UEM College 9 mei, 2006
Dinsdag - donderdag
Experimentatie 2 personen aanwezig steeds 25-30 minuten per proefpersoon Jullie krijgen een tekst protocol Just drinks
Nieuw in dit experimentje:Logsqare wordt gebruikt voor video capture, in activity-
maps
Procedure1. De NFC-lijst2. De planningstaken (korte versie experiment (3 taken)3. Questionnaire (drastisch ingekort, alleen opinions, op
papier)
UEM College 9 mei, 2006
Data (stap 1)
1. CSV-file van NFC scores2. CSV-file van taken (raw data)3. Ingevulde korte (papieren) eind-
vragenlijst 4. Videos en met activity chart
afbeelding
UEM College 9 mei, 2006
Data (stap 2)
Runnen macros
Raw data > preciezere data
Fill up the gaps > mbv. videos
Preciezere data > SPSS
UEM College 9 mei, 2006
11121314151617181920
21222324252627282920
31323334353637383930
UEM College 9 mei, 2006
Data (stap 3) & verwerking
Voeg data NFC scores en Vragenlijst toe aan SPSS file Jullie krijgen van mij nog instructies tzt. Welke variabelen ik
erin wil hebben Stuur ALLE databestanden naar mij per mail Ik krijg dus per proefpersoon:
Nfc-csv PET-csv Pet-csv.xls (preciezer) Pet csv.xls.final (nog preciezer,no gaps) SPSS-bestand compleet (alles!)
Alleen de videos en activity charts kunnen op de Usabilitylab PC blijven staan. Zet dit duidelijk in en folder met groepnummer, daarin mapjes met proefpersoon nummer/naam
UEM College 9 mei, 2006
En dan….analyse!
Jullie krijgen dan zo snel mogelijk van mij een samengevoegd spss-bestand terug (19-20 mei)
26 mei, 23:59 (een week later) is de deadline voor het groeps rapport
Wees precies!! Je kunt tussendoor altijd mailen als er een probleem
is
SUCCES!!