I am doing some research, non-academic and just for fun, into something where I believe the thing under consideration is being more concentrated among a population with time. This sort of feels like a Pareto Chart but I need it to also display over time.
I’d rather not give my example use case, and I don’t even have the dataset yet but a few hypothetical examples:
I have the alcohol consumption in ounces for 1000 people over a ten year period. Assume the data reflects that as a population fewer people are consuming more of the overall alcohol consumption. How could I visualize this?
Suppose I am evaluating 100 non-profits and looking at their donations received. I categorize the donations as being Very Large, Large, Medium, Small, Very Small. Over a 10 year period fewer and fewer non-profits are receiving what I would consider Very Large and Large donations. Is there way to chart this phenomena?
The more I write the more I think the answer may to decide on an arbitrary cutoff (say 80%) of all alcohol consumption or Very Large/Large donations and see how many in my population that covers. Seeing how the 80% included population varies over time could indicate the level of concentration (i.e., if 80% includes fewer and fewer of the population then it’s clearly becoming more concentrated).
Sorry for the stream of consciousness. Does any of what I wrote make sense? Any advice or examples where you have seen this analysis performed? Thanks!
Or you could set a cutoff on percentage of population (e.g. top 10% by alcohol consumption) and see how their average consumption increased, compared to average of the entire population (e.g. the top 10% alcohol consumers consume twice as much as 10 years ago, while the bottom 90% consume the same amount). Or how their share of total alcohol consumption changed over time (e.g. top 10% alcohol consumers consume 50% of all alcohol now vs. 30% 10 years ago).
Here’s an example that compares income of top 10% vs. income of the bottom 90%, which I think is analogous to what you’re trying to do.
In that graph, the Huns become more and more dominant over time relative to the other groups. That’s an example with a fixed Y-axis (% of 100, relative to one another), but you can also do it with a limitless/absolute max Y (like in my examples).
Thank you everyone for the advice. First, scr4, you helped me think about the problem differently. Instead of how many does top 10% include, what does 10% earn which feels more intuitive.
Second, Reply, I think that would show my data well. Unfortunately, after processing my data today it doesn’t actually show what I thought it would. The trend actually seems to indicate becoming more distributed with time rather than more concentrated. I need to look over my algorithms again and apply some of the visualization methods shown here to be sure. Thanks again.
Oh, if you’re just exploring the data and not trying to present it yet, a bubble chart could help too (size of bubbles being another dimension on top of xy, and color of the bubbles labeling which bucket it belongs to). They’re good for interactive explorations of comparative growth.
The trend actually seems to indicate becoming more distributed with time rather than more concentrated. I need to look over my algorithms again and apply some of the visualization methods shown here to be sure. Thanks again.