Restricted abortion access between 2017 and 2020 experienced increases in abortion rates.
Design Decisions and Rationale:
-
Data Transformation
-
Honesty: 0.5
For this graph, we intentionally selected data from an external source, limiting it to states that enacted restrictions between January 2017 and November 2020 to produce a biased dataset. By selectively choosing what information is shown, we are able to manipulate the data to better suite our proposition.
Source: JAMA Internal Medicine, 2021
-
Honesty: 0.5
-
Regression Line
-
Honesty: 1
This regression line was created from the cherry-picked data to confuse the viewer. While the line was made honestly, the purpose of it was to deceive the viewers into believing that there was an increase in abortion rate despite these states having stricter abortion laws.
-
Honesty: 1
-
Title
-
Honesty: -2
The title is deceptive as it leads the viewer into the conclusion that the rising abortion rate is due to the stricter abortion laws, when that is not the case.
-
Honesty: -2
-
Color
-
Honesty: -1
The color red was chosen deliberately to emphasize the increasing line, making it seem more dramatic. Its brightness makes it difficult for the viewer to overlook it when looking at the graph.
-
Honesty: -1
-
Sizing
-
Honesty: -1
The sizing was made to be smaller to show the large spacing between each of the points, exaggerating the graph. Additionally, this sizing makes the regression line more dramatic and eye-catching to the viewer.
-
Honesty: -1
Design Decisions and Rationale:
-
Data Transformation
-
Honesty: 0.5
This graph has all of the data with no transformations. All numbers were directly pulled from the data with no additions or subtractions.
-
Honesty: 0.5
-
Lack of Regression Line
-
Honesty: 1
Unlike the previous graph, this one does not include a regression line. This was done purposely to allow the viewer to see the whole variability of the data. Specifically, the large number of points centered around the origin of the x-axis and the origin. Adding a regression line would only add more confusion and not help lead the viewers to the conclusion that we are attempting to prove.
-
Honesty: 1
-
Title
-
Honesty: 2
The title is fairly honest for the majority of the data; however, it downplays the outliers and fails to acknowledge that some states have increased access. This title was tailored to fit the purpose of our proposition, and also does not acknowledge that other factors are involved with the abortion rate changes.
-
Honesty: 2
-
Sizing
-
Honesty: 2
The sizing of the axes was created to include all of the points, including the outliers. It makes it seem like the dots were closer to the x-axis than if the scales were adjusted to be closer.
-
Honesty: 2
-
Point Design (Shape, Color)
-
Honesty: 1.5
We changed the shape of our data to be open circles so it is easier to see points with similar values compared to closed circles. This allows viewers to see how much overlap there is around the origin and lean further towards the idea that there was not much increase in the abortion change. As for the color scheme, we stuck to neutral colors so that viewers could focus on all of the points.
-
Honesty: 1.5
When creating our visualizations, we decided that presenting our data through scatterplots was straightforward since it was already clean and easy to work with. The Plotly library made it simple to visualize a relationship between clinical access and abortion rate changes between different states. However, it was challenging to figure out how to present the data in a way that supported our specific propositions and made the trends appear more convincing. The scatterplots were our first attempt at visualizing the data, but we wanted to experiment with bar plots as well. However, it did not work out the way we wanted to. All the states were in the x-axis, and while it was easier to interpret the bars by comparing their lengths, having all states listed would only increase cognitive load, and we thought that the audience may spend more time and difficulty trying to understand our visualization, so we went back to using scatterplots. The biggest surprise was how much the title and a single fitted line shaped first impressions. Small choices such as scale, color, and whether to show a line changed what viewers believed before they inspected the points. We also made basic accessibility choices by using color-blind-safe defaults and points shaped as open-circles so that overlaps are visible without relying on color alone. Our audience is a general college classroom, so we favored simple encodings, neutral colors, and captions that reduce misreadings.
After creating our visualizations, we now define “ethical analysis and visualization” to be presenting data earnestly while also acknowledging context that was omitted. The design choices we made regarding titles, regression line, and axes ranges can guide audience interpretation, but maintaining ethics requires us to be transparent about our choices. We believe that transparency also draws the line between “acceptable” persuasive choices and “misleading” ones since highlighting a real trend is not only more acceptable, but also encourages the audience to explore the visualization compared to omitting data, which is more misleading because the audience may believe a proposed conclusion at face value.
However, there are limits to what these charts can claim. Clinic counts do not capture telehealth, travel across state lines, population change, or policy timing, so we cannot make causal claims from these plots. Framing matters because titles, fitted lines, and axis ranges can pull attention toward a story the data does not fully support. Persuasive choices are acceptable when they highlight real patterns in the data, but they become misleading when they imply causal conclusions the evidence cannot justify.If we were to improve our visualization, we would add a small multiples view by region or an interactive brush so readers can test subsets of the data for themselves rather than rely only on our framing.