Of the 1,248 mail in ballots with problems, 969 are ones that couldn't be sent to the voter

About 8-9% of the 1 million registered voters in Philly (about 80-90,000) request mail-in ballots. As the ballots start going out, or the filled-in ones start coming in, the City Commissioner's office starts publishing an Excel file of those that have issues. That list says the name, voting Ward, age and ballot problem reason. As of September 30, 2026, that list has 1,248 names on it.

Of that 1,248 list of names, 969 of those names are voters who never got a ballot at all. USPS tried to deliver it, couldn't and returned the ballot to the city city’s mailing as undeliverable. The city then flags that ballot in SURE, Pennsylvania's voter system.

The other 279 are the mail-in failures you normally hear about. 151 arrived without a signature on the declaration envelope, 79 with an ID verification issue and 49 with no secrecy envelope. Note that these are pre-opening issues... If you voted for a fake candidate (Darth Vader), or your writing or checkmark was illegible, that's an issue we won't know till the votes are opened and counted.

The deficiencies list split into two groups, with distinct age characteristics

The median “undeliverable” ballot belongs to a 41-year-old. The median age of every other voter on the list is in the 70s: 73 for no signature, 75 for ID not verified, and 74 for no secrecy envelope.

There’s a clear monotonic correlation too, the younger you are, the more likely your ballot was returned as undeliverable. For voters under 25, 97% of their deficiencies were undeliverable. For voters age 25-34, it was 96%. For voters 85 and over, it was 47%.

This age-correlated pattern is not surprising: younger voters move more (for a variety of reasons, jobs, family situations, housing instability, etc.) so are more likely to have address changes that result in undeliverable ballots. By comparison, older voters are likely to have more issues in filling out government forms, and/or having outdated IDs.

Several areas are disproportionately negatively affected by mail-in ballot rejections. Interestingly, they tend to be the wealthier areas of Philly

One of Philadelphia's wealthiest neighborhoods, Rittenhouse, has more flagged voters than any other ZIP, 102 of them, or about 8% of the total. Graduate Hospital, another well-to-do area, is the next most affected at 90. There are 46 zip codes in Philadelphia, and yet the top 10 account for 49% of all the rejected ballots.

The obvious interpretation of this data is that Rittenhouse people are younger, and therefore move more, which explains the greater undeliverability issues. However, compare this to Center City zip 19107, which has about 40% less people yet has 70% fewer mail deliverability issues. Or even more starkly University City (19104) has a population of 55,000, 2x the population of Rittenhouse 19103 at 26,000 and even younger - and yet has half the deliverability issues, a ratio differential of 4x. This implies a causal mechanism distinct from the younger-people-move-more hypothesis above.

By ward it is more concentrated still. Ward 8 has 140 flagged voters, more than any other in the city, Ward 5 has 81 and Ward 21 has 67. Twenty-five of the city’s 66 wards have ten or fewer, and the median ward has 13 as of the September 30 data.

Ballots with issues, excluding deliverability issues, by ZIP and income

Let's strip out the 969 undeliverable ballots for a second, and look only at the 279 ballots with issues like bad signatures. Rittenhouse drops from 1st place to 9th as the most affected zip.

Now here's the surprising part (to me at least), there is no income relationship. Contrary to what the media portrays, lower income zips, on average, per 1000 residents, are not more likely to be disparately impacted by bad signatures, bad voter IDs, etc.

As I was doing this analysis, I started with the assumption that rejected ballots were mostly going to be from poorer areas, as that's the way media tends to portray the disparate impacts of tighter voter and ballot verification rules. So I was surprised to find that, no, in fact, in Philadelphia mail-in ballots at least, amongst ballots actually received, there is no discernible correlation.

Now, let’s look at whether there’s a pattern between Undeliverability and Income

Here the income pattern is very clear - the higher the income, the higher the likelihood of undeliverable ballots. In zips where households make $40,000 per year, there's about 0.5 undeliverable ballots per 1000 residents, whereas in zips with incomes $100,000, that number is about 1.8 per 1000, a 3.6x increase. The correlation is 0.72 and statistically very strong. Note that this 3.6x increase is only partially explained away by higher voting rates in richer zips (a differential of about 2x in Philly).

This is surprising because there should be significantly more housing and family instability at lower incomes, which should cause address shifts more often, resulting in more undeliverable ballots.

Looking at the high income end of these graphs, we see two types of high income neighborhoods. First, we see Rittenhouse and Center City again: 20-something professionals who switch jobs and apartments every 2 years, so it's not surprising that their ballots are undeliverable. But then we also see neighborhoods like Fairmount and Society Hill with high rates of undeliverability, despite low housing turnover rates. This contradicts the prior hypothesis that high ballot undeliverability is driven largely by young professionals moving around.

Conversely, looking at the low income end of the above two graphs, you see very low undeliverability for lower income zips. This is surprising as there is likely to be significantly higher housing turnover in these zips, due to family structure volatility, evictions, and lower rates of homeownership.

The point being that, even controlling for the higher voting rates (a differential of 2x), higher income zips tend to have more non-deliverability issues with ballots, even amongst distinct subgroups of income (young professional vs settled families or rich retirees).

Suggested next steps for the Voting Commission and/or election experts to go deeper

Notify people the day the city knows via phone. Many voters register their phone numbers as part of their voter registration (I did as part of a recent update), so call them, don't just mail a notice. This is particularly important for lower-income, less housing-stable voters, who are also the ones most likely to have moved without updating the registration address that their ballot was mailed to.

Match names before throwing a ballot out. National research shows ID-not-verified ballots and 151 missing signatures are usually clerical errors, either by the voter or by the government in the system. In the latter case, it is a high burden for a typical voter to try and fix a government system's error. I'm personally in a 6-month long process to do this with the PA Treasury for an address with a spelling error on it!

The age-defined pattern for undeliverability vs other issues means different remedies are needed for distinct failure modes. Younger voters never received their ballots in the first place, whereas older voters ballots are more likely to be rejected due to deficiencies. Divergences in deficiency reasons mean distinct root causes, which require distinct curative actions.

Notes, Sources, and Methodology

Source data: City Commissioner’s deficiency list last updated Sept. 30, 2026, pulled Oct. 2, 2026. The definition of undeliverable comes from the Pennsylvania Department of State’s mail-in ballot guidance, version 3.4.1, page 7. That section was added in the September 18, 2026 revision specifically to cover undeliverable ballots.

This analysis focuses entirely on the mail in ballot deficiencies as an absolute figure, not as a function of total mail in ballots received, which is an area for potential future investigation.