Designing with Intent: A Framework for Evaluating Affective Goals in Visualizations
“Data visualizations are not neutral.” This contentious but widely acknowledged assertion forms one of the key arguments of Lee-Robbins and Adar (2023) in their paper, “Affective Learning Objectives for Communicative Visualizations.” With reference to Bloom’s framework, the paper identifies three broad domains of learning objectives, i.e. cognitive, affective, and psychomotor.
Using learning objectives as a framework to describe and assess the communicative intent of visualizations, the authors aimed to create a principled way of comparing and assessing designs and enabling external critique of visualizations. The resulting framework is also intended to aid designers identify and state the affective intent of their visualizations described as “goals regarding an audience’s reaction or response to appraisals, attitudes, or values.”
As revealed in the paper, designers tend to associate their work with cognitive learning objectives, even though most visualizations serve affective ends. This is because affective learning objectives are challenging to measure and are stigmatized due to perceived violations of neutrality. However, the authors noted that affective intents are usually made explicit in visualizations intended for advocacy and social justice.
In developing the referenced framework, the authors revised Bloom’s affective taxonomy which separates learning into three domains: cognitive, affective, and psychomotor. In Bloom’s Taxonomy, affective verbs span a continuum: progressing from the lowest level, where the viewer will receive or ‘perceive,’ to ‘respond,’ ‘value’, ‘believes,’ and terminates at the peak of ‘behaves’ where the viewer fully internalizes the values. In the revised taxonomy by the authors, affective intents/verbs start at observe the range of beliefs, then to position oneself within that range of beliefs, then to strengthen one’s belief, then to connect and compare several beliefs with each other, and finally to behave consistent with that belief. The authors went ahead to conceptualize affective noun dimensions as comprising appraisal, attitude, value, and value orientation. Emotion was not included in the taxonomy because the authors consider it not an end but a means to an end.
The authors guided designers to state the learning objectives of their visualizations in a semi-structured interview involving 12 participants. All participants created at least one affective learning objective. All but one also created a cognitive learning objective. Participants were invited to self-report if they considered the learning objectives of the adapted Bloom’s Taxonomy a good fit for their intent. Majority said yes, while one each indicated respectively that only the affective objectives fit, and only the cognitive objectives fit.
Varying opinions were reported by designers on whether they thought their designs directly affected their audience. Some were unsure of how to assess if their visualization was successful in their affective goals. Others indicated that they are not aware of their affective intent. However, the majority agreed that even though they had affective intents, the individual differences within the audience would moderate each person’s reaction to their visualization.
As a proof of concept, the authors used the new taxonomy to evaluate an affective visualization titled “Juneteenth: an Inflection Point in the Struggle for Freedom.”
As shown above, the use of curved timelines was to make the viewer observe the tardy progression of the timeline to freedom for black people from 1619 to 1866. The descriptive text was intended to make the viewer accept the narrative that Abraham Lincoln “was not an abolitionist.” Rather, he used emancipation as a military strategy against the popular belief that he did it on moral grounds. It was also intended for value orientation i.e. to make the user re-orient their value system to accept equity as desirable.
Two limitations of the adapted taxonomy were observed by the authors: its inapplicability to data art, and personal projects. The study participants indicated that learning objectives would be helpful in the following areas: big or novel projects, communicating with clients or a team, evaluating designs, and choosing amongst design options. Participants further submitted that they were likely to use learning objectives at the planning stages to provide clarity to the design process.
The authors conclude that their work offers a pathway for designers to think about their visualizations from both affective and cognitive lenses, contrast alternatives relative to specific articulation of their goals, and assess their work. They invite future work to expand on the idea of measuring and evaluating the success of affective visualizations.
Reference
E. Lee-Robbins and E. Adar, “Affective Learning Objectives for Communicative Visualizations,” in IEEE Transactions on Visualization and Computer Graphics, vol. 29, no. 1, pp. 1–11, Jan. 2023, doi: 10.1109/TVCG.2022.3209500.
