ljw1004
569
11
3
I don't like "relative change graphs". This graph shows how income (blue) and rent (orange) have changed over the years in Honolulu County, Hawaii. Data: https://github.com/arilamstein/censusdis-streamlit/blob/main/county_data.csv
The three graphs all pick different years as the "zero" baseline: 2005, 2010 or 2020. It's the exact same data, just with a different baseline. And the three graphs tell different narratives!
(2005) "Rents have been consistently and significantly higher than income"
(2010) "Rents and income have risen pretty much in lockstep"
(2020) "Rents and income have risen pretty much in lockstep, except for those lucky years in 2005-2009 when rent was crazy cheap."
I think it's bad practice to pick a visualization where (1) the interpretation of the data will change drastically based on an arbitrary choice of how to plot, (2) the arbitrariness of that choice is usually lost on viewers, and often not even noticed by the plotter.