GeoTUI::Results - Study 2

Back 06 March 2007
Experiments at the IFP (Prototype version 3)

small_s3010039.jpg
GUI / Mouse		 small_s3010017.jpg
TUI / 1-puck prop		 small_s3010018.jpg
TUI / 2-puck prop		 small_s3010019.jpg
TUI / ruler prop

  Cutting planes performed

[GUI / Mouse]

[TUI / 1-puck prop]

[TUI / 2-puck prop]

[TUI / Ruler prop]
 Scoring of cutting planes performed: [ 1 2 3 4 ]
  1. Exact
  2. Nearly exact
  3. Quite far
  4. Absurd
 Caption :
  • Green : border of the map.
  • Red : cutting line expected.
  • Blue : cutting line performed by the user.
  Times to perform a cutting line
Times of selection for first exercise
Exercise 1


Times of selection for first exercise
Exercise 2


Averages times for both exercises
Average for exercises 1 and 2

Box plots and average times per interaction to select a cutting line.
The performances for the four conditions of interaction differed significantly:
(one-way analysis of variance)
  • In the first exercise: F3,44=19.18, p<.001
  • In the second exercise: F3,44=4.96, p<.001
  • In both exercises: F3,92=11.79, p<.001

Results of significance of the differences between the averages:

Significance by paired t test
  Positioning of the hands
Distribution of manipulations
Exercises 1 and 2

Distribution of one-handed manipulations (left and right hand) and two-handed manipulations of props, and percent of subjects who kept, at least once, one hand on the validation pad during a series of several cutting planes.
  Number of cutting lines performed
Averages numbers of cutting planes performed
Exercises 1 and 2

Averages numbers of cutting planes performed for each exercise, using different interactions.
 Results of significance:
Significance by paired t test
  Calculation of significance
* One-way analysis of variance were performed using the Analysis Tools of MS Excel.

* Paired t tests and Wilcoxon Signed Rank tests were performed using the following web calculators:
t-Value and p-Value: http://faculty.vassar.edu/lowry/t_corr_stats.html
z-Value and p-Value: http://faculty.vassar.edu/lowry/wilcoxon.html

* Paired t tests were also performed using the following formula in a spreadsheet to calculate the t-Value:
(Extracted from: http://mathworld.wolfram.com/Pairedt-Test.html)

Given two paired sets Xi and Yi of n measured values.
X^_ and Y^_ are the average of Xi and Yi, respectively.

Let X^^_i=(X_i-X^_)
and Y^^_i=(Y_i-Y^_)

Define t by
t==(X^_-Y^_)sqrt((n(n-1))/(sum_(i==1)^(n)(X^^_i-Y^^_i)^2))

This statistic has n-1 degrees of freedom.

and then the associated p-Value was obtained using the following web calculator: http://www.danielsoper.com/statcalc/calc08.aspx

Back Top
Last modification : 03 April 2007,
Webmaster : g [dot] riviere [at] estia [dot] fr
URL : http://www.estia.fr/~geotui/experiments2/
Valid HTML 4.01 Strict   Valid CSS!