6-2 Eye-tracking Data - Fixation Analysis
The log-based and questionnaire could provide both subjective and objective analysis of users’ learning performance on our system. However, the users’ pattern of depicting the visualizations could not be observed through the system log. Eye movement data could support such analysis and acted as another objective measurement of how users engage in the visualization.
To further investigate the effect of format, task and goal orientation on the participants’ eye movement behavior, we specify the following model to estimate the total fixation duration time in seconds and fixation count. Same with previous section, the linear regression model was used for estimation.
Total fixation duration time
𝐹𝐹𝐿𝐿𝑃𝑃𝐿𝐿𝑂𝑂𝐿𝐿𝐿𝐿𝐿𝐿𝐹𝐹𝐸𝐸𝐿𝐿𝐿𝐿𝑂𝑂𝐿𝐿𝐿𝐿𝐿𝐿𝑇𝑇𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖 = 𝛽𝛽0+ ∑2𝑗𝑗=1𝛽𝛽𝑗𝑗𝐺𝐺𝐿𝐿𝐿𝐿𝑙𝑙𝑂𝑂𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑂𝑂𝐿𝐿𝑂𝑂𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖+ ∑4𝑗𝑗=3𝛽𝛽𝑗𝑗𝐹𝐹𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑂𝑂𝑖𝑖 +
The equation (4) test how goal orientation impact the total fixation duration time.
Also, it test the influences of the format of graph visualization and the type of learning tasks on the total fixation duration time. The index i stands for each observation in our integrated collected data. The effects of dummy variable GoalOrientation on fixation duration are denoted by the parameters 𝛽𝛽1 𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽2, which represents the difference level
‧
of the high degree observation and the low degree observation of the goal orientation, and the difference level of the high degree observation and the middle degree of the goal orientation. 𝛽𝛽3 𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽4 are coefficients for the effect of dummy variable Format on fixation duration, which represents the difference level of the line format and the bar format, and the difference level of the radar format and the bar format. 𝛽𝛽5 reflects the effect of dummy variable Task on fixation duration, which represents the difference level between search fact task and inference generation task on fixation duration.
𝛽𝛽6, 𝛽𝛽7, 𝛽𝛽8, 𝛽𝛽9, 𝛽𝛽10, 𝛽𝛽11, 𝛽𝛽12 are coefficients for the effect of control variable.
𝛽𝛽6, 𝛽𝛽7 𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽8 are estimate of dummy variable AOI on fixation duration, which represents the difference level of the AOI Question, AOI Title, AOI Visualization and the AOI Legend as the baseline. 𝛽𝛽9 is estimate of dummy variable ProgrammingExperienced, which represents the difference level between the participants experienced in programming and the beginner. 𝛽𝛽10 𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽11 are estimate of dummy variable ExamNumber, which represents the difference level of the exam 2, exam 3 and the exam 1 as the baseline. 𝛽𝛽12 is estimate of dummy variable Gender, which reflects the effect of difference level between the male and female. Finally the 𝜀𝜀𝑖𝑖 is the level-1 error term.
In sum, the effect of goal orientation on fixation duration time was found between high and middle degree of the goal orientation at the level of 0.01. The effect of format was found between radar and bar at the level of 0.05. The results show goal orientation and format has impact on the total fixation duration time. The estimated results for total fixation duration time in ALL page (without considering AOIs) are summarized in Table 6.4. Also we have found significant difference of control variables AOI between AOI Visualization, AOI Question, and the baseline. Thus we further estimate the same regression model on AOI Visualization and AOI Question. The results are summarized in Table 6.5-6.6.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 6.4 Estimated results for total fixation duration time in All page.
Estimate Std. Error t value Pr(>|t|)
Constant 5.178 2.917 1.775 0.07643 *
Goal orientation (Low) -1.405 1.787 -0.786 0.43218 Goal orientation (Middle) -4.506 1.631 -2.763 0.00592 ***
Format (Line) 0.634 1.581 0.401 0.68858
Format (Radar) 3.948 1.562 2.527 0.01177 **
Task (SearchFact) -0.398 1.274 -0.312 0.75487
AOI (Question) 26.271 1.758 14.942 < 2e-16 ***
AOI (Title) -2.667 2.065 -1.292 0.19697
AOI (Visualization) 19.791 1.758 11.256 < 2e-16 ***
Programming-experienced 4.538 1.769 2.565 0.01058 **
Exam number (2) -11.040 1.545 -7.148 2.87e-12 ***
Exam number (3) -12.441 1.554 -8.005 7.30e-15 ***
Gender (Female) 3.035 1.294 2.345 0.01940 **
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.
‧
Table 6.5 Estimated results for total fixation duration time in AOI Visualization.
Estimate Std. Error t value Pr(>|t|)
Constant 30.4693 5.1134 5.959 1.60e-08 ***
Goal orientation (Low) -3.8283 3.4788 -1.100 0.2728 Goal orientation (Middle) -6.0958 3.1475 -1.937 0.0546 *
Format (Line) -0.2194 2.9898 -0.073 0.9416
Format (Radar) 7.6618 2.9841 2.568 0.0112 **
Task (SearchFact) -2.4405 2.4365 -1.002 0.3181
Programming-experienced 6.7904 3.3503 2.027 0.0444 **
Exam number (2) -21.7752 2.9841 -7.297 1.35e-11 ***
Exam number (3) -20.9936 2.9898 -7.022 6.13e-11 ***
Gender (Female) 4.1293 2.4657 1.675 0.0960 .
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.
Table 6.6 Estimated results for total fixation duration time in AOI Question.
Estimate Std. Error t value Pr(>|t|)
Constant 33.696 6.581 5.120 8.77e-07 ***
Goal orientation (Low) -1.860 4.477 -0.415 0.678384 Goal orientation (Middle) -8.227 4.051 -2.031 0.043935 **
Format (Line) -1.054 3.848 -0.274 0.784430
Format (Radar) 3.223 3.841 0.839 0.402613
Task (SearchFact) 1.689 3.136 0.539 0.590867
Programming-experienced 6.100 4.312 1.415 0.159116
Exam number (2) -13.916 3.841 -3.623 0.000391 ***
Exam number (3) -18.984 3.848 -4.934 2.03e-06 ***
Gender (Female) 5.726 3.173 1.804 0.073072 *
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.
The equation (5) test how goal orientation, format and task impact the fixation count. The index i stands for each observation in our integrated collected data. The
‧
effects of dummy variable GoalOrientation on fixation duration are denoted by the parameters 𝛽𝛽1𝐿𝐿𝐿𝐿𝑃𝑃𝛽𝛽2, which represents the difference level of the high degree observation and the low degree observation of the goal orientation, and the difference level of the high degree observation and the middle degree of the goal orientation.
𝛽𝛽3𝐿𝐿𝐿𝐿𝑃𝑃𝛽𝛽4 are coefficients for the effect of dummy variable Format on fixation count, which represents the difference level of the line format and the bar format, and the difference level of the radar format and the bar format. 𝛽𝛽5 reflects the effect of dummy variable Task on fixation count, which represents the difference level between search fact task and inference generation task. 𝛽𝛽6, 𝛽𝛽7, 𝛽𝛽8, 𝛽𝛽9, 𝛽𝛽10, 𝛽𝛽11, 𝛽𝛽12 are coefficients for the effect of control variable. 𝛽𝛽6, 𝛽𝛽7𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽8 are estimate of dummy variable AOI on fixation duration, which represents the difference level of the AOI Question, AOI Title, AOI Visualization and the AOI Legend as the baseline.𝛽𝛽9 is estimate of dummy variable ProgrammingExperienced, which represents the difference level between the participants experienced in programming and the beginner. 𝛽𝛽10𝐿𝐿𝐿𝐿𝑃𝑃 𝛽𝛽11 are estimate of dummy variable ExamNumber, which represents the difference level of the exam 2, exam 3 and the exam 1 as the baseline. 𝛽𝛽12 is estimate of dummy variable Gender, which reflects the effect of difference level between the male and female. Finally the 𝜀𝜀𝑖𝑖 is the level-1 error term.
In sum, the results are similar to the results of fixation duration time. The effect of goal orientation on fixation count was found between high and middle degree of the goal orientation at the level of 0.05. The effect of format was found between radar and bar at the level of 0.05. The results show goal orientation and format has impact on the total fixation duration time. The estimated results for total fixation count are summarized in Table 6.7. We also found significant difference of control variables AOI between AOI Visualization, AOI Question, and the baseline. Thus we further
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
estimate the same regression model on AOI Visualization and AOI Question. The results are summarized in Table 6.8-6.9.
Table 6.7 Estimated results for total fixation count in All page .
Estimate Std. Error t value Pr(>|t|)
Constant 34.548 13.208 2.616 0.00915 ***
Goal orientation (Low) -5.762 8.090 -0.712 0.47660 Goal orientation (Middle) -14.642 7.383 -1.983 0.04786 **
Format (Line) 3.893 7.159 0.544 0.58685
Format (Radar) 14.155 7.072 2.001 0.04584 **
Task (SearchFact) -8.163 5.768 -1.415 0.15758
AOI (Question) 148.581 7.961 18.665 < 2e-16 ***
AOI (Title) -13.764 9.349 -1.472 0.14154
AOI (Visualization) 107.028 7.961 13.445 < 2e-16 ***
Programming-experienced 21.319 8.010 2.662 0.00801 ***
Exam number (2) -59.067 6.993 -8.446 2.78e-16 ***
Exam number (3) -70.921 7.037 -10.079 < 2e-16 ***
Gender (Female) 12.994 5.860 2.217 0.02701 *
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.
‧
Table 6.8 Estimated results for total fixation count in AOI Visualization.
Estimate Std. Error t value Pr(>|t|)
Constant 165.999 23.792 6.977 7.82e-11 ***
Goal orientation (Low) -13.322 16.187 -0.823 0.4117 Goal orientation (Middle) -18.860 14.645 -1.288 0.1997
Format (Line) 1.913 13.912 0.138 0.8908
Format (Radar) 24.357 13.885 1.754 0.0813 .
Task (SearchFact) -11.667 11.337 -1.029 0.3050
Programming-experienced 27.244 15.589 1.748 0.0825 .
Exam number (2) -112.821 13.885 -8.125 1.21e-13 ***
Exam number (3) -113.876 13.912 -8.186 8.50e-14 ***
Gender (Female) 24.906 11.473 2.171 0.0314 **
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.
Table 6.9 Estimated results for total fixation count in AOI Question.
Estimate Std. Error t value Pr(>|t|)
Constant 203.829 28.301 7.202 2.27e-11 ***
Goal orientation (Low) -9.669 19.254 -0.502 0.6163 Goal orientation (Middle) -26.157 17.420 -1.501 0.1352
Format (Line) -5.494 16.548 -0.332 0.7403
Format (Radar) 12.339 16.516 0.747 0.4561
Task (SearchFact) -11.536 13.485 -0.855 0.3936
Programming-experienced 35.153 18.542 1.896 0.0598 *
Exam number (2) -80.804 16.516 -4.892 2.43e-06 ***
Exam number (3) -115.565 16.548 -6.984 7.53e-11 ***
Gender (Female) 17.431 13.647 1.277 0.2034
* Significance (Sig.) at 0.1 level, ** Sig. at 0.05 level, *** Sig. at 0.01 level.