Skip to main content

Table 1 Mechanical Turk user study analysis

From: Unboxing cluster heatmaps

Type Task Number Question text Nodes Type Mean χ 2 df p-value  
Training 1 1 Is the highlighted cluster mostly positive or mostly negative? Clusters N/A  
Training 1 2 What is the height of the tree? N/A N/A  
Timed 1 3 Which of the highlighted elements has the highest value? Leaves Score 0.380 4.578 5 4.695 E-01  
Timed 1 4 Is the highlighted cluster mostly positive or mostly negative? Clusters Score 0.661 14.650 5 1.197 E-02
Timed 1 5 What is the height of the tree? N/A Error 3.083 23.278 5 2.987 E-04
Training 2 6 Which of the highlighted elements is furthest away from the root? Leaves N/A  
Training 2 7 Which of the highlighted elements are siblings? Leaves N/A  
Timed 2 8 Which of the highlighted elements is furthest away from the root? Leaves Score 0.605 63.540 5 2.250 E-12
Timed 2 9 Which of the highlighted elements is furthest away from the root? Clusters Score 0.732 14.705 5 1.170 E-02
Timed 2 10 Which of the highlighted elements are siblings? Clusters Score 0.864 21.662 5 6.070 E-04
Training 3 11 Which two of the highlighted elements are more closely clustered? Clusters N/A  
Training 3 12 How many visually distinct clusters do you see in this visualization? N/A N/A  
Timed 3 13 Which two of the highlighted elements are more closely clustered? Siblings Score 0.738 29.499 5 1.850 E-05
Timed 3 14 Which two of the highlighted elements are more closely clustered? Leaves Score 0.275 12.775 5 2.558 E-02
Timed 3 15 Which two of the highlighted elements are more closely clustered? Clusters Score 0.352 9.539 5 8.941 E-02 ·
Timed 3 16 Which of the highlighted elements is least similar to its neighbors? Clusters Score 0.283 13.726 5 1.745 E-02
Timed 3 17 How many visually distinct clusters do you see in this visualization? N/A Value 8.794 31.138 5 8.796 E-06
  1. Shows the type of question (training or timed), the task set, question number and text, the node type of the choices (leaf nodes, sibling nodes, or cluster nodes), the value type (score, absolute error, or raw value), overall average, and results (χ 2-test statistic, degrees of freedom, and p-value) from the per-question Kruskal-Wallis tests by technique. See Figs. 4 and 5 for the distribution of values for these questions broken down by technique
  2. Legend: *** p≤ 0.001, ** p≤ 0.01, * p≤ 0.05, . p≤ 0.1