1. Diagrams and Text in Instruction: Comprehension of the Assembly Process Julie Heiser Marie-Paule Daniel Ginet Barbara Tversky Special thanks to Christina Vincent!

2. Why assembly? - A process that requires a match between visual and verbal internal representations and the external counterpart. - Requires action- structure into function. - A common task for all ages…also a common problem.

3. Assembling a BBQ

4. BBQ assembly broken into steps

5. Why assembly? - A process that requires a match between visual and verbal internal representations and the external counterpart. - Requires action- structure into function. - A common task Why furniture? - Has both structure and function - Requires nearly perfect action on mental representations, perhaps driven by instructions. - A common experience, becoming more common with assemble-your- own everything. - Nearly universal. - Also a common problem (sample of responses).

6. A simple 2 drawer dresser…

7. Assembly instruction project outline: Experiment 1: Collection of instruction protocols Experiment 2: Compilation of individual protocols into a ‘mega-description’ Experiment 3: Quality rating of instruction protocols Experiment 4: Efficiency and effectiveness evaluation of representative instructions

8. Experiment 1 - 42 undergraduate Introductory Psychology students - Between subjects, 21 S’s in 2 conditions 1) Experience questionnaire 2) Spatial Ability tasks - Mental rotation test - Money navigation test 3) Assemble TV stand (w/only picture of completed stand) 4) Write instructions for assembly - unconstrained - constrained

9. Spatial ability: Mental Rotation Test (Vandenburg & Kuse, 1978)

10. Spatial ability: Money Task (Money & Kuse, 1966)

11. Experiment 1- method - 42 undergraduate Introductory Psychology students - Between subjects, 21 S’s in 2 conditions 1) Experience questionnaire 2) Spatial Ability tasks - Mental rotation test - Money navigation test 3) Assemble TV stand (w/only picture of completed stand) 4) Write instructions for assembly - unconstrained - constrained

12. Given parts and a picture of the completed stand…...….….Assemble

13. Experiment 1- method - 42 undergraduate Introductory Psychology students - Between subjects, 21 S’s in 2 conditions 1) Experience questionnaire 2) Spatial Ability tasks - Mental rotation test - Money navigation test 3) Assemble TV stand (w/only picture of completed stand) 4) Write instructions for assembly - unconstrained (2 pages) - constrained (1/2 pg- minimal amount of information)

14. Constrained Unconstrained TV Stand Instruction Protocols

15. Example 1: unconstrained, low spatial ability

16. Example 2: constrained, high spatial

17. Example 3: unconstrained, low spatial

18. Protocol Analysis Text - effect of space constraint - effect of drawings (condition 3) Drawings - diagrammatic elements - types of diagrams (interactive, structural, etc.) - individual differences

19. Text Analysis Effect of space constraint: - Strong decrease of number of propositions: 43.33 vs. 21.60 - Resistance of assembly action category - Decrease of part description, other non-assembly categories Effect of presence of drawings on text: - Overall, no significant effect on the number of propositions: 38.57 vs. 42.33 - Number of propositions referring to actions is less with drawings (9.3 vs. 6.7) - When drawings are present, less time indicators present in text. - Information in text duplicates information in drawings.

20. Diagram Analysis Condition (U vs C)  Time to assemble  Number of steps specified  Step elements  Start protocol with text or diagram?  Number of separate parts drawn  Number of interactive drawings  Number of non-interactive drawings  Parts labeled, how?  Elements in diagrams (lines, arrows)  Diagram representations  End result or procedural diagrams  Integrated or exploded  Quality of drawing  Quality of 3-D Mental Rotations score Money task score Assembly experience Self-rated assembly ability Self-rated mechanical ability

21. Diagram Analysis Condition (U vs C)  Time to assemble  Number of steps specified  Step elements  Start protocol with text or diagram?  Number of separate parts drawn  Number of interactive drawings  Number of non-interactive drawings  Parts labeled, how?  Elements in diagrams (lines, arrows)  Diagram representations  End result or procedural diagrams  Integrated or exploded  Quality of drawing  Quality of 3-D Mental Rotations score Money task score Assembly experience Self rated assembly ability Self rated mechanical ability Independent variables

22. Diagram Analysis Condition (U vs C)  Time to assemble  Number of steps specified  Step elements  Start protocol with text or diagram?  Number of separate parts drawn  Number of interactive drawings  Number of non-interactive drawings  Parts labeled, how?  Elements in diagrams (lines, arrows)  Diagram representations  End result or procedural diagrams  Integrated or exploded  Quality of drawing  Quality of 3-D Mental Rotations score Money task score Assembly experience Self-rated assembly ability Self-rated mechanical ability IndependentDependent

23. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

24. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

25. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

26. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

27. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

28. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

29. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

30. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

31. Unconstrained vs. Constrained - No group differences (MR scores etc., time to assemble) - No difference in # of steps, interactive or structural drawings, - More separate parts drawn in Unconstrained -Diagram representations- dual. -constrained more likely to use new information in diagrams. -Indicating steps- generally used numbers -- unconstrained use more indirect cues - Labeling parts: constrained- part A,part B, etc. unconstrained- top, bottom, side, etc. - Both conditions more likely to start with text than diagram. - Diagram elements- if any, arrows and lines indicate direction or interaction - Integrated diagrams most common. Constrained more likely to use exploded.

32. Individual Differences Condition Unconstrained Constrained Time # of parts drawn # of interactive # structural 9.6 3.43 1.76 1.0 10.9 2.95 1.5 1.2

33. Individual Differences Condition Unconstrained Constrained Time # of parts drawn # of interactive # structural SA Low High Time # of parts drawn # of interactive # structural 9.6 3.43 1.76 1.0 10.9 2.95 1.5 1.2 11.4 3.8.57 1.57 8.1 2.6 2.64.73

34. Correlations MR score and Assembly time = -.531 MR score and Quality of drawings =.543 MR score and Quality of 3-D =.478 Assembly experience and # parts drawn = -.653 MR score and # interactive drawings =.584

35. Assembly instruction project outline: Experiment 1: Collection of instruction protocols Experiment 2 (analysis in progress): Compilation of individual protocols into a ‘mega-description’ Experiment 3: Quality rating of instruction protocols Experiment 4: Efficiency and effectiveness evaluation of representative instructions

39. Assembly instruction project outline: Experiment 1: Collection of instruction protocols Experiment 2: Compilation of individual protocols into a ‘mega-description’ Experiment 3 (experiment in progress): Quality rating of instruction protocols Experiment 4 (to be continued…): Efficiency and effectiveness evaluation of representative instructions

40. (Preliminary) general conclusions - People have different ideas of what makes effective manuals. Is this in production of instruction or in comprehension, or both? - There are performance differences in assembly tasks depending on prior experience and spatial ability. - There are individual differences in internal mental representations for external representations in assembly tasks. -Important to design instructions using design principles developed by cognitive psychologists. -Would be great to have automated visual instructions that can balance text and diagrams in relation with the experience of the user!