Using ArcGIS Pro 3.4, I'm trying to use Spatial Join to identify which counties are "within or partially within" a congressional district, but exclude those that merely touch a boundary. For example, Chaves county (yellow polygon) should be in multiple districts (black boundary lines), while Deaf Smith should be in only one district.
Using "largest overlap" doesn't work because it only returns the single largest overlap and "intersect" doesn't work because shared boundaries create many incorrect duplicates.
SHPs available here - https://www.census.gov/geographies/mapping-files/time-series/geo/cartographic-boundary.html
Join Features is supposed to accomplish this using "Overlap" but is still returning polygons with shared boundaries. (The bottom image is directly from this link and shows that shared boundaries should NOT be included.)
This is the result of Join Features using "Overlaps" and one-to-many, four counties returned rather than just the one:
Have you tried utilizing a [small] negative search distance? I've successfully used this before to exclude bordering features. Since these are large county boundaries, you can probably get away with a larger negative value than if you were dealing with a lot of small features.
perhaps have a look at the Select by location options, there are some cases not covered by spatial join.
Select By Location graphic examples—ArcGIS Pro | Documentation
I think you've got some slithers of overlaps from the datasets. Certainly recommend Jack's approach, and possibly Tabulate Intersection (Analysis)—ArcGIS Pro | Documentation might help you to figure out what's going on with the potential overlaps - or may be used to achieve the join instead (remove <1% overlap etc).
The data is directly from public sources and they should be topologically aligned. Do folks have examples of datasets where the Join Features "Overlaps" relationship works as described?