In the '21-'22 NBA season, Steph Curry shot better than Jayson Tatum on 2-pointers (52.7% vs 52.4%) and 3-pointers (38% vs 35.3%) but when we consider both of these types of shots together, Steph hit at a worse overall rate on field goals than Tatum (43.7% vs 45.3%). This is an example of Simpson's paradox.
With the specialization of players' roles in the NBA and differences in shot selection and shooting abilities, there are several examples of Simpson's paradox In the following graphic, the edges between certain nodes represent such instances. In the graphic, I took the top 25 scorers and plotted each of their
1. 2-pointer field goal percentage (2P%, x-axis),
2. 3-pointer field goal percentage (3P%, y-axis),
3. Overall field goal percentage (FG%, color of the markers),
4. Relative proportion of field goal attempts that are 2-pointers (size of markers).
Finally, when there's an example of Simpson's paradox, I plotted
5. Line segments connecting the pairs of players whose 2P%, 3P%, and FG% satisfy Simpson's paradox
One thing you can notice is that a necessary (but not sufficient) condition for a pair of players Player A (better overall FG%) and Player B (better 2P% and 3P%) to satisfy the "paradox'' is that Player B must be above and to the right of Player A on the graph, but Player A has a larger and darker marker, since Player A shoots more 2s overall. This is under the (not true for all players but true here) assumption that all players shoot 2-pointers at a higher percentage than 3-pointers.
The property is also transitive, as exhibited with DeMar DeRozan, Joel Embiid, and Zach Lavine, and also by Dejounte Murray, Anthony Edwards, and Steph Curry.
This analysis isn't meant to judge any players as better than others; it's simply an interesting phenomenon that occurs between the best scorers in the league because of their different ways of scoring. Feel free to click around below to focus on certain parts. Double-clicking resets it. Special shoutout to Sports Reference (SR), specifically Basketball Reference (BR), for all of the data on the players. Enjoy!