Does Sketching Skill Relate to Good Design

1
DETC2005-85499
Does Sketching Skill Relate to Good Design?
Prof. Maria C. Yang Industrial Sys.
Engineering University of Southern California
Dr. Jorge G. Cham Mechanical Engineering Californi
a Institute of Technology
2
Sketching in Design

• Sketches important for capture communication of
  ideas Ullman, et al 90 Verstijnen, et al., 98
• Sketching linked with design cognition Nagai
  Noguchi 03 Suwa and Tversky 97 Goel 95
• Research context
• What about sketching skill?
• What motivates a designer to sketch?
• Sketching ability linked to design quality?
• Sketching necessary for effective design?
  Schutze et al 03

3
Sketch assessment

• Sketch survey
• mechanical recall
• drawing facility
• visualization of new objects
• Correlate with
• Sketch fluency (logbooks)
• Design outcome

4
Related work

• Visual representation in design McKim 80 Schön
  Wiggins 92
• Sketching is dialogue Cross Shah, et al
  Goldschmidt Tovey, et al
• Conceptual design preserves ambiguity Goel
  Kavakli, et al
• Sketching and Design Outcome
• Sketching activity linked to outcome Schütze, et
  al
• 3-D sketching and design outcome Song and
  Agogino
• Dimensioned drawings and design outcome Yang 03
  

5
Testbed

• Sketching ability
• Introduction to Mechanical Design at Caltech
• Month-long final projects (individual or teams)
• PHOTO OF AN ME71 PROJECT HERE

6
Research questions

• Is sketching ability generic?
• Haves vs. Have nots
• Is sketching ability linked to sketch fluency?
• Is sketch ability linked to engineering design
  performance?
• What is the effect of sketch instruction?

7
Survey to assess drawing skill(do try this at
home)

• In 3 minutes, draw a bicycle with as much detail
  as possible.
• Hold out the items given to you in your
  non-dominant hand (left-hand for right-handed
  persons). In 3 minutes, make a drawing of your
  hand and the items (items given were two small
  candy bars).
• Visualize and draw the following in 2 minutes A
  rectangular box that is open at the top. Inside
  the box is a rubber ball. The front of the box
  has a large button, and each side of the box has
  a large X painted on it.

8
Goal of the survey

• Mechanical recall Bike task
• Ability to recall and sketch familiar,
  non-trivial mechanical objects
• Drawing facility. Hand task
• Ability to create realistic, well-composed
  drawings
• Requires little imagination or visualization
• Novel Visualization.Box task
• Ability to visualize objects that consist of
  specific features.
• Decouple different skills for translating
  mechanical ideas into sketches

9
Assessing sketches

• Sketching skills
• Challenge is subjectivity of good or a bad
  sketch
• Clear scoring criterion for each task (1 to 5)
• Mechanical recall
• Demonstrate grasp of structure and operation of a
  bicycle (Does this look like a bicycle and could
  it be ridden?).
• Drawing facility
• Proportions of hand and realism (Does this look
  like a hand?)
• Novel visualization
• Proportions of box, accuracy of 3D perspective,
  and realism

10
Representative sketches
Level 1
Level 3
Level 5
Level 1
Level 3
Level 5
Mechanical Recall Task
Drawing Facility Task
Novel Visualization Task
11
Design outcomes

• Sketch fluency
• Relatively objective yet informative
• Paper design logbooks
• Grades for final project
• Final project grades assessed by the instructor
  (not authors)
• Rankings by external judges
• Independent professional design engineers
• Spearman Correlation between parameters and
  outcome

Rs correlation coefficient N sample size
12
Results Discussion

• Sketching ability
• No correlation between bike task and hand or box
  tasks.
• Doing well in one task does not correlate with
  doing well in others -gt skills appear to be
  independent

13
Sketching ability and sketch fluency

• Bike task
• Uncorrelated with total sketches
• Neg. correlated with 3D -gt Good mech recall, less
  likely to draw 3D
• Hand and box tasks
• Correlate positively with total sketches
  perspective sketches
• Possible conclusions
• Sketch quantity linked to drawing ability (Hand)
  need (low Bike)
• Good drawing more likely to sketch, but only if
  poor at mech. visualizing
• Good mech. visualization but poor at drawing -gt
  work out designs in their head (dont sketch at
  all)

14
Sketching and Design Outcome

• Little correlation bet. fluency OR ability and
  design outcome
• Quality of design does not related to sketch
  activity, or facility with drawing
• Possible conclusions
• Design thinking not always reflected in sketches
  (or other rep.)
• Sketching result of need for visual
  representation
• Sketching a behavioral output, based prefs or
  abilities.
• Neither being a good sketcher (good artist) or
  having good mechanical sense (gear head) meant
  a good design

15
Role of sketch instruction

• Sketch instruction 14 students also enrolled in
  CAD course that introduced sketching
• Is instruction in sketching linked to sketch
  quantity or design outcome?
• Avg drawings higher with instruction, but not
  stat. sig. (large var, in qty of sketches, w/inst
  started out higher)
• Perspective drawings reflect sketching skill
• Sketch instruction may result in higher sketching
  output

16
Conclusions

• Question 1 Is sketching ability generic?
• Sketching skills not created equal. Good at the
  hand or box task not good at bike
• Possible reason Different tasks require
  different cognitive skills
• Question 2 Is sketching ability linked to sketch
  fluency?
• Hand and box task stat. sig. correlated with
  total sketches
• Bike task neg, stat. sig. correlated with 3D
  sketches
• Implication Sketch fluency (partly) determined
  by how much a designer can design without
  drawing.
• Possible reasons
• Good mechanical recall allows visualization of
  designs in ones head (without committing to
  paper)
• Common complaint I dont need to keep a logbook
• Particular sample Caltech engineering undergrads

17
More conclusions

• Question 3 Is sketch ability linked to
  engineering design performance?
• No relationship between any sketch task and
  design outcome
• Good sketchers did not necessarily do well (or
  vice versa)
• Possible reasons
• Engineering design is complex, requires many
  skills sketching is only one
• Question 4 What is the effect of sketch
  instruction?
• Sketch instruction linked to more sketching, but
  no effect on grades or rankings
• Little correlation between total sketches and
  grades
• While sketching can reflect design thinking, it
  is affected by many factors
• Sketching behavioral rather than a necessary
  element of design activity

18
Future work

• Many factors can influence design outcome
• Difficult to tease out confounding variables
• Other potentially relevant factors to
  investigate
• working in a team vs individually
• individuals motivation or personality
• nature of a design task.

19
Implications for engineering design education

• Currently
• US undergraduates instructed in CAD
• Suited to later stages of design
• Teaching sketching by hand is less common
• Flexibility of sketching important in initial
  stages
• Sketching is one way of thinking that some prefer
• Proposals
• Represent and generate design ideas
• Develop visualization skills much as math or
  verbal
• Stanford ME - mandatory course in visual
  thinking
• Caltech ME elective visualization course
• Others?

20
Acknowledgements

• We are very grateful for the generous support and
  guidance of
• Prof. Joel Burdick, California Institute of
  Technology
• Prof. Erik Antonsson, California Institute of
  Technology
• Students of Caltech ME71 and ME170
• Prof. Karl Gröte, Cal State Long Beach
• Dr. Curtis Collins, Jet Propulsion Laboratory

21
A note about outcome consistencies

• Consistency between three metrics of design
  outcome, as measured by the correlation among the
  three metrics. project grade and overall class
  grade was, not surprisingly, statistically
  significant (Rs 0.89). In addition, the
  rankings of the two outside design judges also
  correlated in a statistically significant way
  with each other (Rs 0.59). Interestingly, both
  judges rankings correlated negatively with
  project grade (Rs -0.22), perhaps because
  project grades take into consideration the
  process involved in developing a project, while
  the reviewers rankings are solely based on
  responses to the final artifact itself.