Information
- Publication Type: Master Thesis
- Workgroup(s)/Project(s): not specified
- Date: 2023
- Second Supervisor: Johanna Schmidt
- Open Access: yes
- First Supervisor: Eduard Gröller
- Pages: 131
- Keywords: data visualization, data literacy, processing speed, comprehensibility
Abstract
As static visualizations like diagrams, maps, charts, or drawings get more and more important in everyday life, it is crucial to find out how helpful they actually are, especially in comparison to text. This diploma thesis outlines if visualizations or texts are processed faster by humans and which representation is better comprehensible. For this purpose, we conducted an exploratory study in which we measured processing times and error rates when interpreting either texts or visualizations. In a pre-study, in which each participant had to describe four visualizations in their own words, we found out which parts of a visualization are deemed most important by people. The focus of the participants was on extrema, as well as on certain other values. Furthermore, the participants often compared different values in order to describe the visualization. The pre-study gave us valuable insights which we used for writing the texts for our study. For our exploratory study, where we wanted to find out more about whether visualizations or text can be interpreted more easily, we used 15 visualizations and texts that contained the same information. Each participant had to work on at least six topics, thereof at least three by using visualizations and three by using texts, and had to answer three questions per topic at the end. The time was measured while the participants worked on the topics.We could see that the participants solved topics by using visualizations on average about 1.3 times faster as compared to when they used texts. This difference was statistically significant. We could not find significant differences between the error rates for topics, when participants used visualizations or texts. When texts were used, we found correlations between text lengths and processing speeds, as well as text lengths and error rates. The content of visualizations or texts did not seem to play a role for processing speed or error rates. However, we found cases in which topics with visualizations were solved more than 50 % faster as compared to topics with texts. Our results provide a solid basis for defining further hypotheses regarding the readability of visualizations compared to text. In this thesis, we present the final hypotheses that emerge from our exploratory study. We consider these to be extremely interesting for visualization research, as there is much evidence that visual information can be processed faster than text. Whereby it is worth mentioning that the actual increase in performance may be much lower than often claimed in the media (e.g. `60,000 times faster than text'). Furthermore, our results indicate that no significant differences can be found between visual information and text with regard to the error rate in answering final questions.
Additional Files and Images
Weblinks
BibTeX
@mastersthesis{tuscher-2023-qeo,
title = "Quantitative Evaluation of Reading Times and Error Rates
When Interpreting Visual Content",
author = "Michaela Tuscher",
year = "2023",
abstract = "As static visualizations like diagrams, maps, charts, or
drawings get more and more important in everyday life, it is
crucial to find out how helpful they actually are,
especially in comparison to text. This diploma thesis
outlines if visualizations or texts are processed faster by
humans and which representation is better comprehensible.
For this purpose, we conducted an exploratory study in which
we measured processing times and error rates when
interpreting either texts or visualizations. In a pre-study,
in which each participant had to describe four
visualizations in their own words, we found out which parts
of a visualization are deemed most important by people. The
focus of the participants was on extrema, as well as on
certain other values. Furthermore, the participants often
compared different values in order to describe the
visualization. The pre-study gave us valuable insights which
we used for writing the texts for our study. For our
exploratory study, where we wanted to find out more about
whether visualizations or text can be interpreted more
easily, we used 15 visualizations and texts that contained
the same information. Each participant had to work on at
least six topics, thereof at least three by using
visualizations and three by using texts, and had to answer
three questions per topic at the end. The time was measured
while the participants worked on the topics.We could see
that the participants solved topics by using visualizations
on average about 1.3 times faster as compared to when they
used texts. This difference was statistically significant.
We could not find significant differences between the error
rates for topics, when participants used visualizations or
texts. When texts were used, we found correlations between
text lengths and processing speeds, as well as text lengths
and error rates. The content of visualizations or texts did
not seem to play a role for processing speed or error rates.
However, we found cases in which topics with visualizations
were solved more than 50 % faster as compared to topics with
texts. Our results provide a solid basis for defining
further hypotheses regarding the readability of
visualizations compared to text. In this thesis, we present
the final hypotheses that emerge from our exploratory study.
We consider these to be extremely interesting for
visualization research, as there is much evidence that
visual information can be processed faster than text.
Whereby it is worth mentioning that the actual increase in
performance may be much lower than often claimed in the
media (e.g. `60,000 times faster than text'). Furthermore,
our results indicate that no significant differences can be
found between visual information and text with regard to the
error rate in answering final questions.",
pages = "131",
address = "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
school = "Research Unit of Computer Graphics, Institute of Visual
Computing and Human-Centered Technology, Faculty of
Informatics, TU Wien",
keywords = "data visualization, data literacy, processing speed,
comprehensibility",
URL = "https://www.cg.tuwien.ac.at/research/publications/2023/tuscher-2023-qeo/",
}
thesis