Source: https://www.ers.usda.gov/webdocs/publications/82220/eib-166.pdf?v=42762
This visualization is from a report from USDA (United States Department of Agriculture). The report is about U.S. Trends in Food Availability and Dietary Assessment of Loss-Adjusted Food Availability, 1970-2014. The graph under consideration for this blog, is the 2nd Figure in the report and the first figure under the Findings category. It is a layered donut chart that shows the per person calorie consumption for each of the food category: Fruit and vegetables, Grains, Added fats and oils, Meat, eggs & nuts, Added sugar & sweeteners, and Dairy. The layered donut has two layers, inner layer for the per person calorie consumption for each category in the year 1970 and the outer layer which shows per person calorie consumption for each category in the year 2014 (except for Added fats and oils for which the outer layer shows consumption for the year 2010). Each category is color coded to identify the region demarcations in the donut layers and the numbers within each region indicate the calorie consumed per person for that category.
Things I liked about the visualization:
- Visualization is simple and easy to understand. It depicts what the figure title says clearly, calorie consumption per person for 1970 and 2014 for each food group.
- The graph is well labeled. The calories consumed in each category is clearly visible in contrast to the color of the region for each category.
- The colors chosen are bright and appropriate, there is no overlapping of color shades or similar colors that would make it difficult to understand consumption of each food group.
- The regions are aligned for both years, one in front of the other. This makes it easy to see the difference in calorie consumption for each group for the 2 years.
Things I did not like about the visualization:
- The calorie consumption for Added fats and oils is shown for the year 2010 and not 2014 as data for that category was available till 2010 only. But it has been placed in the outer layer of donut with the rest of categories which depict calorie consumption for 2014, and the title says that outer ring says that outer ring depicts calories by food group for 2014. This is misleading.
- The use of a layered donut chart looks like it is intended to give an idea of the proportion of each food group consumed in both years, with respect to each other. But, there is no indication of the percentage of each food group consumed. One has to try and figure out the proportion of each food group consumed from the relative portions of the donut. This may not give the exact idea of the proportions to the viewers.
- I am not sure I like the idea of layered donuts for comparing the food group consumption for both years. The difference in proportions is not clear by just comparing the concentric rings. The choice of visualization does not do justice to the intent of the visualization.
- Not including the percentage value of each food group may mislead the audience as they may relate an increase in calorie consumed between both years to an increase in proportion of food group consumed, which may not always be true(as in case of Added sugar and sweetners).
Critical Analysis of the visualization:
Let’s analyze the subjective dimension of the visualization:
- Beautiful: I would definitely not call this a very beautiful visualization. The use of a layered donut chart does not seem to be the optimal choice for the purpose. It is not easy to compare the proportion of consumption of one category to the others for one particular year. It is also not easy to compare the change in proportion of food group consumed between the two years. For instance, it is not easy to determine the exact percentage of the food group Dairy’s consumption in year 1970. The use of a stacked bar graph for both years or line graphs would have helped visualizing these details better.
- Functional: The functionality of the visualization could be improved. The current visualization labels only the calories consumed in both years, leaving the viewer to decipher the relative proportion of food group consumed by the size of the donut rings, which is not easy. The functionality could have been improved by separately visualizing the change in relative proportion of food groups consumed with respect to each other, along with the calories consumed. As change in proportion of food group consumed is equally important in deriving any useful insights, it should have been included as well.
- Insightful: The visualization does a fair job in terms of being insightful. The ordinary audience (someone like me) will not necessarily have knowledge of the changing trends in food group consumption and this visualization gives a good idea of how food consumption trends has changed between 1970 and 2014. But this can also be improved. One way of improving could be by including age specific consumption of food groups, which gives a much clearer idea as to which age groups have shifted more from eating healthy food like fruits to consuming more calories of added fats and oils by eating more junk food. The second way is by including information for dietary guidelines for consumption of each food group. This would make the visualization more insightful, as the information on consumption of food groups will now have a context and give more useful insights.
- Enlightening: The entire report is about changing trends in food availability and dietary assessment. But this graph only shows the change in calories consumed among the different food groups. Just by looking at the changing calorie consumption of each food group one cannot initiate any change as there is no clear indication of any impacts of changing calorie consumption on one’s diet/ health. The graph also does not give much information regarding the changing food availability to make any useful decisions regarding the availability of food choices.
Does this graph have a claim?
- No I don’t think so. As mentioned earlier, this graph serves as a blanket visualization for the report which explains the change in trends of food group availability trends from 1970 to 2014. As it is the first graph and is expected to give an overview of all the information that is further explained in the following graphs, this graph needs to give an overview of the intended claim of the report. The report is intended to inform the audience regarding the changing food availability trends and an assessment of the diet of Americans. But by just looking at this graph, we do not see any claims for food availability or diet assessments.
Validation of data:
- We do not know the change in consumption of each category over all the years between 1970 and 2014. Providing the information of just the start and end year may mislead the audience if the trends for the years in between show significant variations, as data can be cherry picked to make it look the way you want. The omission of data on years between the start and end years raises a question on the shown trends, since the results of one particular year can well be an anomaly and thus not indicative of a trend.
Redesign of the visualization:
As I mentioned earlier, the use of a layered donut chart is not the right choice for this visualization. The visualization intends to give an idea of change in calorie consumption per food group per person. I have redesigned the given visualization to give a better understanding of change in calorie consumption between 1970 and 2014. I have also visualized the change in proportion of each food group consumed, to give a clear idea of the change in trends of food group consumption.
Link to redesigned visualization:
References:
Why not to use Pie/Donut charts:
http://geographymaterials.blogspot.com/2015/08/advantages-and-disadvantages-of-pie.html
Using the right chart for comparing values over time:
https://www.cardinalsolutions.com/blog/2016/05/data-visualization-best-practices-part-two-mistakes-to-avoid