CA-30 2028 simulator

IMPUTED Some tracts in Acton, Agua Dulce, Santa Clarita had no 2024 result — they were not in CA-30 before AB 604. Each is placed at its own 2026 standing relative to the district, not at the district average: they ran at about a fifth of the district in 2026, so treating them as average overstated this surface roughly four-fold in exactly the exurban areas that set the Round-1 bar.

Each slider opens at the estimate for the starting point above. Moving one shifts every tract by the change, weighted by how much of that tract the group is — so leaving them alone leaves the observed result untouched.

A slider you move becomes a target, and the model solves for a map that hits it. The sliders you have not touched are readouts: they show what your targets imply for those groups, and they move on their own. Setting Latino support to 30% pulls the renter figure to about 19.6% and the homeowner figure to about 11.8% — not because anyone asked, but because that is who lives in the tracts that had to move.

Within a family nothing moves. White and Latino name people who are never the same voter, so setting one says nothing about the other and the sibling readouts hold still. Between families they overlap and therefore must move: 64.8% of this district's homeowners are White against 57.8% of its renters, so raising White support by 10 points raises the homeowner figure by 6.48 and the renter figure by 5.78. Those are the overlaps, not a rule of thumb.

How much the other families move is measured, not assumed. Each tract carries its own joint composition — what share of it is each combination of ethnicity, tenure and age — fitted to three Census cross-tabulations (B25003 tenure by race, B01001 age by race, B25007 tenure by age) at tract level. Those associations are large: across this district renter rates run 57.9% among Asian householders to 86.8% among Black ones, and 89.9% among under-35 householders to 50.5% among those 65 and over.

Two things are still assumed. The Census publishes the three pairs but no table of all three at once, so the cells carry no three-way interaction beyond the pairs. And two of the three tables count households by a single householder, where this model counts voters: an adult child living in a parent's home is counted at the parent's age and tenure, which overstates how strongly age and tenure move together. Those tables set the shape of the association and never a level — each is rescaled onto this district's own composition first — so nothing in them can move a group's share.

Extreme targets can still be unreachable — under the two unbounded modes a group's average can be met while some tract is pushed past 100%, which is not a share anything can have. The page says so rather than showing you the nearest-looking answer.

Effects remain very unequal in vote terms: White voters are about 60% of this electorate and Black voters about 5%, so the same target shift on the two moves very different numbers of votes. That is not rescaled.

NOT REACHABLE These targets cannot all be met on this district's geography. The projection below is not valid for them — release a target, or move it closer to what the others imply.

ASSUMPTION One consolidated Republican.

The rest of the Democratic field is not modelled — not how many other Democrats run, how strong they are, or how the non-left vote splits. Top-two placement is a question about the whole field, and this tool holds that field fixed.