What the survey actually measured
NerdWallet commissioned YouGov to survey 5,087 U.S. adults across California, Florida, Georgia, New York and Texas in June 2026, with roughly 1,000 respondents per state, weighted to represent each state’s adult population. That’s a reasonable design for state-level snapshots: enough people to say something meaningful about the state as a whole. It is not enough to say something reliable about, say, renters under 30 in Atlanta versus homeowners near Albany — those are exactly the subgroups a 1,000-person state sample struggles to support with precision, because the moment you slice the data further, the number of respondents in any one slice shrinks fast.
The survey also asked about perceptions and self-reported behavior: whether someone has felt stressed about expenses, felt embarrassed about credit card debt, considered moving because of cost. Those are legitimate things to measure — they capture lived experience that a bank statement doesn’t. But they answer a different question than "how large is this household’s debt-to-income ratio" or "did this household miss a loan payment." Feeling behind and being behind can move together, but they aren’t the same measurement, and a survey built to ask about feelings shouldn’t be quietly read as a survey about balance sheets.
Why official surveys build in more caution
It’s worth contrasting this with how the Federal Reserve treats similar terrain. The Fed’s Survey of Consumer Finances warns explicitly that its complex sampling and multiple-imputation methods mean researchers must recalculate standard errors correctly or risk stating false confidence in a number. That’s a technical point, but the underlying lesson generalizes: any survey estimate carries a margin of uncertainty, and that margin widens considerably once you start comparing subgroups rather than a state’s population as a whole.
The Fed’s 2025 Survey of Household Economics and Decisionmaking (SHED) illustrates the second half of the caution. Nationally, the share of adults "doing okay financially or living comfortably" held steady at 73% — a headline of stability. But that stability masked real deterioration among young adults, low-income families, and Black adults. A topline number and a subgroup trend pointed in different directions at the same time. If a national aggregate can hide that much divergence, a five-state survey with roughly 1,000 respondents per state should be read with at least the same humility about what’s happening inside each state’s number.
None of this means the added Fed sources tell us anything directly about California, Florida, Georgia, New York or Texas in this particular survey — they don’t offer a state-by-state rematch. What they offer is a methodological mirror: a reminder of how much can be lost between a topline percentage and the reality underneath it.
A framework for reading the headline
| Claim | Can the survey support it? | Why |
|---|---|---|
| "40% of Georgia adults reported expense-related stress in the past year" | Yes, as a description of the sampled state population | Matches what was actually asked and who was weighted to represent |
| "Georgia households are less financially secure than California households" | No | Self-reported stress isn’t an audited measure of debt, savings, or default risk |
| "This subgroup of Georgia renters under 30 is especially stressed" | Not reliably from this data alone | State-level samples of ~1,000 rarely support fine subgroup breakdowns with confidence |
| "Affordability stress is a widespread theme worth investigating further" | Yes | Consistent broad signal across multiple states in the same wave |
| "Moving to Texas will improve my finances" | No | Survey doesn’t track wages, taxes, insurance, or career effects of relocation |
From a number to a defensible conclusion
The safe path from a survey response to something you can actually act on — intellectually, not literally — runs through a few checkpoints rather than a straight line to a forecast.
flowchart TD A[Sample: who was asked?] --> B[Measure: perception or balance sheet?] B --> C[Aggregation: state average or subgroup?] C --> D[Caveats: size, imputation, distribution] D --> E[Limited, hedged conclusion]
Skip any one step and the "conclusion" at the end is really a guess wearing a statistic’s clothing.
The takeaway, not the verdict
None of this makes the NerdWallet survey useless. Broad, consistent signals — that affordability stress is common across large, diverse states, that credit card debt carries real emotional weight, that a fifth of households across all five states lack a financial plan — are worth noticing precisely because they recur. What the survey cannot do is settle a comparison between states as if it were a scoreboard, or stand in for an individual household’s actual financial condition, or justify a moving van.
Treat the state average as a question, not an answer: what does this number look like once you ask who’s included, what was actually asked, and what a Georgia household making $35,000 in a small city has in common with one in Atlanta earning six figures. Usually, not much — and that gap is exactly what the average was built to hide.