The Story
Registration Target Analysis
Step through the narrativeThe full picture
| Precinct | Municipality | Raw | Eff | Reg. voters | D% | R% | D-R gap | Tier |
|---|
Street-level drilldown
383 special outreach addresses across senior communities
After correcting address parsing (VoteBuilder encodes building numbers as "BLDG" tokens that the voter file doesn't use), most senior-community addresses now match registered voters in the building. The 383 remaining unmatched senior addresses are units with no same-base-address FVE voter at all. Of these, 254 are at Shannondell (Lower Prov 3-4, D+171) and 129 are across Worcester Township's senior facilities (Founders Village, Holly House, Laurel House, Cardinal Crest).
These addresses require institutional outreach - mail campaigns, community-room tabling, and coordination with facility management - rather than ordinary door-knock canvassing.
How we weighted the data
Ben Stevens exported 9,511 "zero registered voter" household addresses from VoteBuilder across 8 Area 4 municipalities. We built a street name normalization crosswalk (VoteBuilder and the PA voter file use different conventions - e.g., "Foxmeadow" vs "Fox Meadow", "5th" vs "Fifth", "Providence Forge" vs "Providence Frg") and cross-referenced every address to classify each into three categories: no active FVE match found (4,599 / 48.4%), ambiguous base match (4,232 / 44.5%), and active conflict (680 / 7.1%). Senior housing (383 remaining confirmed addresses) was separated for institutional outreach.
We then applied canvass-efficiency weights to each address based on its housing profile. Multi-unit addresses receive a modest discount (0.75x) reflecting lower expected contact rates from locked lobbies, higher turnover, and unit-address parsing uncertainty - not vacancy. Streets with 10+ addresses were individually profiled against the voter file for registration density and age demographics.
Data quality note: "No active FVE match" means we could not find an active voter registration at that address using our matching model. It does not mean the address is confirmed unregistered - some fraction are likely data matching failures (different address formatting between systems). During analysis, we discovered and corrected multiple systematic matching artifacts. The effective target count (~3,900) should be treated as an estimate with a confidence range, not a precise count.
| Category | Weight | Example streets |
|---|---|---|
| Residential | 1.0x | Mill Grove Dr, Taylor Way, Brooke Dr |
| Mixed unit | 0.80x | Cameron Ct, Walnut St, W Ridge Pike |
| Occupied apartment | 0.70x | Bayless Pl, Cardin Pl, Roboda Blvd |
| High vacancy apt | 0.35x | Foxmeadow Dr (176 unreg, 38 reg) |
| Vacant apt | 0.25x | Foxmeadow Cir (119 unreg, 0 reg) |
| Senior housing | Special | Shannondell Dr, Holly House |
Data: VoteBuilder export (April 26, 2026) cross-referenced against PA Full Voter Export (June 8, 2026 vintage).