No mark is an island: Precision and category repulsion biases in data reproductions

McColeman, Caitlyn; Harrison, Lane; Feng, Mingyue; Franconeri, Steven

doi:10.1109/tvcg.2020.3030345

Cited by 10 publications

(24 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also find that the value of a mark (the reference value) has a more powerful effect on reproduction than the channel chosen: tall bars are more biased than small areas, even though position (bar) is one of the least biased channels overall. The context of the reference value also shows strong bias (e.g., angle), similar to past work where participants tend to be more biased and less precise with a value further from the ends of the range [34,52] (e.g., "edge effects"); it also aligns with psychophysical observations [26], where low and high ends of the data range can serve as perceptual anchors (e.g., "the angle is 10 • from 90 • is perceptually congruent). Alternatively, for a channel like position (bar), participants perform better around the median value, possibly because they resort to near-mean estimations when their memory falters, which would also be consistent with better performance around the mean of The number of marks (6 is the interpolation of the model).…”

Section: The Context Of a Visualizationsupporting

confidence: 80%

“…More recent work [77,87] identified similar impairments. In other reproduction tasks, like the one used in the present study, surrounding values in a display created memory biases, such that recollections of a single relevant value were repulsed from the 0, .5, and 1.0 proportion of a second larger reference bar [52]. Memory bias has been shown even for values presented alone, such that tall bars with a high height:width ratio were underestimated, and wide bars with a low height:width ratio were overestimated [14].…”

Section: Context and Bias Effects On Visual Judgementsmentioning

confidence: 89%

“…Bias Bias and precision precision are different facets for the same set of responses. Participants could be precise but consistently underestimate (or overestimate) the value [17,52]. They could be imprecise but generally right on average.…”

Section: Methodsmentioning

confidence: 97%

“…The short duration exposure, along with unlabeled axes, prevent participants from recoding stimuli into other forms [51] and suppress top-down effects like prior knowledge. The duration is adequate for testing visual working memory [51] and provides ample time for the vision system to encode information (e.g., comparing correlation in scatterplots [67], estimating two-value ratio in bar charts [52], etc.). The inclusion of a blank screen as a mask and a different redrawing location together eliminated visual aftereffects.…”

Section: Methodsmentioning

confidence: 99%

“…Concurrently considering multiple factors and their conflicts and interactions would likely be difficult for designers, such that an automated model might be important for weighting their complexities. We were inspired by the recent modeling work [20,31,40,43,75] and appealed to psychophysical laws [31,52,69,82], entropy [15,68,70,71], perceptual proxies [39,58], serial-position and ordering effects [36], visual memory (e.g., [3,10,54]), neighborhood effects [87], and distractors [77]. We are capable of providing (very) preliminary recommendations given the inputs (see Fig.…”

Section: Implications and Future Directionsmentioning

confidence: 99%

See 4 more Smart Citations

Rethinking the ranks of visual channels

McColeman¹,

Yang²,

Brady³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Data can be visually represented using visual channels like position, length or luminance. An existing ranking of these visual channels is based on how accurately participants could report the ratio between two depicted values. There is an assumption that this ranking should hold for different tasks and for different numbers of marks. However, there is surprisingly little existing work that tests this assumption, especially given that visually computing ratios is relatively unimportant in real-world visualizations, compared to seeing, remembering, and comparing trends and motifs, across displays that almost universally depict more than two values.To simulate the information extracted from a glance at a visualization, we instead asked participants to immediately reproduce a set of values from memory after they were shown the visualization. These values could be shown in a bar graph (position (bar)), line graph (position (line)), heat map (luminance), bubble chart (area), misaligned bar graph (length), or `wind map' (angle). With a Bayesian multilevel modeling approach, we show how the rank positions of visual channels shift across different numbers of marks (2, 4 or 8) and for bias, precision, and error measures. The ranking did not hold, even for reproductions of only 2 marks, and the new probabilistic ranking was highly inconsistent for reproductions of different numbers of marks. Other factors besides channel choice had an order of magnitude more influence on performance, such as the number of values in the series (e.g., more marks led to larger errors), or the value of each mark (e.g., small values were systematically overestimated). Every visual channel was worse for displays with 8 marks than 4, consistent with established limits on visual memory. These results point to the need for a body of empirical studies that move beyond two-value ratio judgments as a baseline for reliably ranking the quality of a visual channel, including testing new tasks (detection of trends or motifs), timescales (immediate computation, or later comparison), and the number of values (from a handful, to thousands)

show abstract

Section: The Context Of a Visualizationsupporting

confidence: 80%

Section: Context and Bias Effects On Visual Judgementsmentioning

confidence: 89%

Section: Methodsmentioning

confidence: 97%

Section: Methodsmentioning

confidence: 99%

Section: Implications and Future Directionsmentioning

confidence: 99%

See 3 more Smart Citations

Rethinking the ranks of visual channels

McColeman¹,

Yang²,

Brady³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Truth or Square: Aspect Ratio Biases Recall of Position Encodings

Ceja¹,

McColeman²,

Xiong

et al. 2021

IEEE Trans. Visual. Comput. Graphics

Self Cite

View full text Add to dashboard Cite

Fig. 1. The Mekko chart at left uses height to encode relative market share, similar to a bar chart. But the present results suggest that the aspect ratio of each mark may bias this height judgment. When asked to reproduce the vertical position of a single bar mark in a bar chart, bars with wide aspect ratios were overestimated, bars with tall ratios were underestimated, and bars with square ratios showed no systematic bias. This pattern of bias appeared within memory, suggesting that value comparisons that occur across time and space (e.g., bars in separate graphs) would most likely be distorted.

show abstract