Visual Search Efficiency for Features in Chernoff Faces

A Systematic Review of Experimental Studies on Data Glyphs

“…Some study results indicate that increasing the number of data dimensions affects the performance of glyph designs negatively [24], [47], [58], [63] with position encodings (linear and circular profiles) being more robust compared to color encodings in high-density situations [24]. As with the number of dimensions, there is evidence that performance drops with increasing the amount of visible glyphs on the screen [42], [46], [63], [79], [81], [82], [83]. This seems like a logical conclusion due to the required additional effort in visual search involving a higher number of entities.…”

“…The amount of glyphs visible to participants changed from 5-50 [79]; 5-15 [82], [83]; 6-18 [46]; 9-23 [81]; 4-300 [42]; and 30-48 [63]. In all seven studies participants were affected negatively by an increasing number of data points, as we discuss in Section 4.6.2.…”

“…Yet, in all studies the performance dropped with an increasing number of data points independent from the mapping of data to face characteristics. Based on this outcome Siva and co-authors concluded that participants performed a serial search and were not able to pre-attentively identify the stimulus [82], [83]. Therefore, the perception of abstract data glyph faces compared to human faces was shown to be different.…”

“…The goal of most of the studies with varying number of glyphs was to investigate changes in performance when increasing the number of visible data points in grid layouts [79], [82], [83], geographic maps [42], [81], and node-link diagrams [63]. The amount of glyphs visible to participants changed from 5-50 [79]; 5-15 [82], [83]; 6-18 [46]; 9-23 [81]; 4-300 [42]; and 30-48 [63].…”

“…The glyphs used in these experiments were either faces [79], [81], [82], [83], unique glyph designs (i.e., MILSTD2525 [46], arrow glyphs [42]), or star glyphs [63].…”