Now I would like to create a parent map that shows these several districts as a complete township map.
Is creating the districts and grouping them into a township the best approach for this use case? I don’t want to go too far with our 9 districts yet if this is the wrong approach.
Warrington township already exists as relation 3521479. Relations are the usual way to record administrative and political boundaries in OSM as the larger ones tend to have too many nodes to be easily editable as part of a single way and the use of smaller segments allows the boundary ways themselves to be referenced multiple times by the relations for the neighbouring areas and also by the higher level entities (where relevant).
I think voting areas would normally be recorded as boundary=political rather than boundary=administrative as has been used for the township. Where recorded these would also normally consist of ways that have been split where three or more areas meet to allow the shared geometry to be reused. In some cases they may also re-use existing features but this is normally discouraged if the legal division is expected to stay the same regardless of e.g. a river moving over time.
For more information on boundary tagging in general see this wiki page.
Note that there may also be import guidelines to be followed or licencing issues to be considered if the source of the information isn’t a ground survey.
It is normally not desired in OSM that electoral districts are entered. These are pure import data, which are not OTG verifiable. If the data is publicly available, then it can be used without import. For example by means of “uMap”. This has already been discussed several times.
The data for at least one boundary=political have already been entered, but they are poorly tagged: they do not contain this tag, but likely should. However, I’d agree with @pyram that the value of these being in OSM is quite low. It isn’t that these aren’t entered (they are), but as was said (and implied), these are always going to be imported from some other geo database, so use those data directly, there isn’t any real need to put them into OSM. (And OSM is not the first place I or many people would look to find these data).
Yes, boundary=administrative data (with an associated tag of admin_level=*) are useful data to be in OSM, by wide consensus. But boundary=political data? Not so much. It’s done, yes, but why, really? Especially as these data are so spottily-entered into OSM and are always found elsewhere besides OSM.
TBH, I have found myself interested in elections where I wish the political boundaries had been in OSM as the maps in the news coverage were either terrible or non-existent.
I definitely know what you mean by that, as we have a fair number of transit agencies and even television stations (e.g. their weather broadcasts) that use OSM data, right on-screen in their local news coverage. (Do they properly give attribution to OSM? Sometimes yes, sometimes no).
See Santa Clara County, California - OpenStreetMap Wiki where “more than a few” of our local media and transit agencies see OSM data and say “hey, let’s use THOSE data!” And then they do. It’s pretty cool to see a mountain ridge you’ve added, or maybe a small village that would otherwise go unmapped suddenly appear on your 11 PM weather forecast!
I haven’t seen local political boundaries in OSM used for election coverage, but you might, so, as I said “it is done,” but it isn’t widespread. What I’d say about this is “add to OSM what you think might be (or actually is) useful in your area.” You never know when some downstream user (television station, transit agency…) might also find OSM data useful. In fact, I have found (over many years, even decades) that “official” data sources, like statewide public area databases that delineate park boundaries, nature reserves, etc. will incorporate OSM data, and these will “watch each other” over many years, improving in both directions. This takes time, and may or may not happen, but I have seen it happen before.