As a professional data vizzer , I can honestly say I’ve never seen a stacked bar chart and gone oh I can I have a conversation with it. I’m always stuck teasing out if this category grew or if it’s bc the category under it is larger than before. Whatever secrets it holds it holds it well. I welcome anything else.
I like your "trimmed grey bar" showing the total for each category but two comments:
- I would put the numbers of the individual INSIDE each segment (probably with a white font) to reduce confusion as to which bar the number should be associated with.
- Since the width is not a variable, I would choose a much narrower width as long as the number fits in it. That way, the overall grey bar will be much thinner.
In the end, it goes back to what you want to visualise. The relative dimensions or the absolute dimensions. There will always be one that is more important than the other.
I like this attempt quite a bit, although it feels like it's a bit victim to the issue that pie charts have -- the additional "width" dimension throws off the comparison. If we have more and more subcomponents, it becomes harder and harder to use. The benefit of the stacked graph is it keeps everything in the same dimension as the comparison.
I agree that stacked bars are mostly unnecessarily difficult, and adding a total behind a set of normal bars is one way to improve the situation.
In my book (Data visualization for biomedical scientists: creating graphs and tables that work. Amsterdam: VU University Press; 2022.) of which you have a copy, I spend some pages on discussing alternatives (pp 119-122). For instance, I developed a combination of lines and bars to depict the number of covid cases overall, and by continent. I'll email it to you separately, then perhaps you can attach it to this post.
As a professional data vizzer , I can honestly say I’ve never seen a stacked bar chart and gone oh I can I have a conversation with it. I’m always stuck teasing out if this category grew or if it’s bc the category under it is larger than before. Whatever secrets it holds it holds it well. I welcome anything else.
I like your "trimmed grey bar" showing the total for each category but two comments:
- I would put the numbers of the individual INSIDE each segment (probably with a white font) to reduce confusion as to which bar the number should be associated with.
- Since the width is not a variable, I would choose a much narrower width as long as the number fits in it. That way, the overall grey bar will be much thinner.
In the end, it goes back to what you want to visualise. The relative dimensions or the absolute dimensions. There will always be one that is more important than the other.
Also who wouldn’t love you and your posts??? Love mom
Thanks Jon I delete that from my list.
I like this attempt quite a bit, although it feels like it's a bit victim to the issue that pie charts have -- the additional "width" dimension throws off the comparison. If we have more and more subcomponents, it becomes harder and harder to use. The benefit of the stacked graph is it keeps everything in the same dimension as the comparison.
Hi Jon,
I agree that stacked bars are mostly unnecessarily difficult, and adding a total behind a set of normal bars is one way to improve the situation.
In my book (Data visualization for biomedical scientists: creating graphs and tables that work. Amsterdam: VU University Press; 2022.) of which you have a copy, I spend some pages on discussing alternatives (pp 119-122). For instance, I developed a combination of lines and bars to depict the number of covid cases overall, and by continent. I'll email it to you separately, then perhaps you can attach it to this post.
Cheers,
Maarten Boers