Is Miami a failure of the assimilation model?
In a recent piece, Jason Richwine of the Center for Immigration Studies argued that Miami is failing on assimilation metrics. What does the data show?
What is Richwine's argument?
Richwine's analysis found that aggregate Spanish-at-home went from 37% in 1980 to 59% in 2000 to 67% in 2024. Among native-born Miami residents age 20-29, he reports that Spanish-at-home went up from 11% in 1980 to 49% in 2024; English-only-at-home in comparison was 42% in 2024. Richwine sees this as a failure of a commonly held assimilation model, and concludes the predicted intergenerational decline this model often sees as inevitable "has apparently not even started." All six of Richwine's numbers replicate cleanly against IPUMS USA microdata (chart at right).
Richwine uses Miami-Dade County. Adding Broward and Palm Beach counties (the standard "Miami metro area") drops the rate by
What's measured here
Share of Miami residents who speak Spanish at home, by age group and birthplace. "Native-born" includes US-born and Puerto Rico-born respondents (consistent with Richwine's universe). Note that the rising aggregate Spanish-at-home rate is not driven by a falling native-born share: the native-born share of Miami-Dade residents has held steady at 47-49% across the 2000, 2010, 2020, and 2024 ACS samples.
The two "Miamis"
Miami-Dade County is the city's home county. The Miami-Fort Lauderdale-West Palm Beach metro area adds Broward and Palm Beach — both with much smaller Hispanic populations. Geographic definition changes the headline rate by about 24 points.
Source
IPUMS USA microdata: 1980 decennial census + American Community Survey 2000 and 2024 (5-year sample for 2024).
Spanish retention in Miami is genuinely exceptional. Most other major Hispanic metros show declining rates of retention over time.
The cleanest test of whether Miami's language pattern represents real exceptionalism is what happens by the third generation. Each panel below tracks the share of Hispanic 3rd+-generation young adults speaking Spanish at home in one metro from 1980 to 2024. The national average has collapsed from 55% in 1980 to 25% in 2024. Despite this, many major Hispanic-receiving metros show flat or modest decline (the national fall reflects 3rd+-gen Hispanics who live outside concentrated metros), while Miami remains both distinctly high and distinctly flat.
Why this matters
The third-plus generation is a critical test for assimilation because it shows what happens once an immigrant community has fully integrated into the local environment.
How I identify "3rd+-generation"
The modern census doesn't ask where your parents were born, so I identify 3rd+-generation people by finding young adults (age 20-35) who still live with a native-born parent. If both the young adult and the parent on the same census form were born in a US state (BPL < 100), they're 3rd+-generation. Puerto Rico-born respondents are excluded from "native" for this language metric since Puerto Rico is itself a Spanish-speaking jurisdiction, classifying PR-born grandparents as native would conflate "US-experience" with "Spanish-environment-experience" and bias the test toward finding Spanish retention. (Note: this differs from the deck's structural metrics — BA+, LFPR, wages, homeownership, etc. — where PR-born ARE counted as native because the Spanish-vs-English distinction isn't germane to those outcomes.)
For cross-survey validation, I looked at CPS (which has direct parent-birthplace fields and does not require co-residence to determine generation) and found that roughly 66% of true 3rd+-gen Hispanic 20-35-year-olds in Miami co-reside with a parent, so this rule captures about two-thirds of the true population. The bias is largely uniform across metros, and Miami's strict-rule sample is closer to its broader US-born Hispanic 20-35 universe than other metros' (Miami selection effect = −19pp vs comparison-metro average −32pp), so the cross-metro ranking holds.
Where lines are missing
El Paso shows data only from 2020 onward, not because there are few 3rd+-generation Mexican-American adults in El Paso (a deeply established population in absolute terms), but because IPUMS USA's METAREA variable, which the pipeline uses to identify metros in the 1980-2010 5% samples, doesn't reliably code El Paso residents. The pipeline switches to PUMA-level reconstruction for the 2020 and 2024 ACS samples, which restores El Paso to full coverage and accurately reports its 3rd+-gen Spanish retention. San Antonio is missing only one cell (the year 2000) because that specific (metro × year × strict 3rd+-gen) combination dropped below the 30-unweighted-respondent suppression threshold. Cells with fewer than 30 unweighted respondents are suppressed throughout; Miami's earlier-year cells qualify but have wider uncertainty than later years.
Source
IPUMS USA: 1980, 1990, 2000 5% decennial census + 2010, 2020, 2024 American Community Survey 5-year samples.
Miami's pattern is unusual even among other modern concentrated immigrant groups.
Holding Miami Hispanics against the major modern Asian-American immigrant communities under matched conditions (same dataset, same outcome, same generation methodology) the classic intergenerational shift to English shows up clearly. Filipino-American (
How retention is calculated
Universe: ACS respondents age 20-35 in each (group × geography) cell. Numerator: PERWT-weighted count of respondents who report speaking the group's ancestral language at home (LANGUAGE = Spanish for Hispanic; Mandarin/Cantonese for Chinese; Japanese for Japanese; Tagalog/Cebuano/Ilocano for Filipino). Denominator: PERWT-weighted total of the cell. Cells with fewer than 30 unweighted respondents are suppressed.
How groups are identified
Asian groups by detailed race code. Hispanic by the standard ACS Hispanic-origin question. Generation logic on Slide 6: 1st-gen = foreign-born; 2nd and 3rd+-gen via household structure (young adult still living with a parent → read the parent's birthplace from the same record).
Within-Miami variation by Hispanic origin
Among reportable Miami-Dade 3rd+-gen cells (age 20-35, n ≥ 30, both observable parents US-state-born): Cuban 65% (n=322), South American (other) 69% (n=58), Other Hispanic 62% (n=32), Puerto Rican 59% (n=65). Mexican (n=15), Dominican (n=18), and Central American (other) (n=24) cells are too small to report. All reportable cells sit well above the Asian-American comparators.
Asian groups not shown at 3rd+-gen
Korean, Vietnamese, and Asian Indian mass immigration is post-1965, so most 3rd+-gen members are still under 20. National-pool Korean/Vietnamese/Asian Indian 3rd+-gen cells exist for a handful of metros but represent the leading-edge grandchildren of the very earliest arrivals — too thin to anchor a generational trajectory.
Outside Miami and the Texas border, 3rd-generation Spanish-at-home retention is low — and it has been falling everywhere for 40 years.
The two snapshot maps below show the decline of 3rd-gen+ Spanish retention geographically. Texas, New Mexico, and Arizona (the historic Mexican-American Southwest) collapsed from 70-88% to 20-33% over 44 years, while Florida, New York, and other Caribbean-immigration destinations sit much closer to where they started. As seen in the other slides, Miami's resilience is real, but it is not representative of the national trajectory.
Universe and identification
3rd+-generation = US-state-born Hispanic adult age 20-35, child of head, with BOTH the household head and head's spouse (where present) US-state-born (BPL < 100), confirmed via a household-structure join. This is the same strict rule used on Slides 3, 5, and 7 — applied here at the state level instead of the metro level. Spanish-at-home = ACS/Census LANGUAGE field = 12. PERWT-weighted rates.
Sample-size tiers
3rd+-gen sample sizes are very uneven across states. Wilson 95% CIs at the typical p≈0.30 rate: n=30 gives ±16pp, n=50 ±13pp, n=100 ±9pp, n=200 ±6pp. I use a hard floor of n ≥ 50 (gray fill, no rate shown), a "thin sample" tier at n = 50–99 (muted color, thin border, ±10–13pp CI), and full color at n ≥ 100. Smaller states such as Vermont, South Dakota, and DC drop out in every year; the southeastern Mountain West drops out in earlier years.
Map projection and geography
Albers-USA projection (lower 48 + repositioned Alaska/Hawaii). State boundaries from US Census Bureau 2020 cartographic boundary files (us-atlas 10m). Puerto Rico is not shown — the 3rd+-gen universe definition requires US-state birth, so PR-born respondents are 1st-gen by construction. Earlier years (1980, 1990) intentionally look sparse: small states had thin Hispanic populations then and the strict 3rd+-gen identification produces n < 50 in many places.
Speaking Spanish at home isn't the opposite of speaking English well — and Miami's exceptionalism emerges specifically at the 3rd+ generation, not earlier.
Richwine treats Spanish-at-home rates as evidence against assimilation. There's some merit to this, but the joint distribution of home language and English fluency tells a different story. Among all 3rd+-gen Miami Hispanics 20-35, 61% are bilingual (Spanish at home + English well/very well) and only an estimated 3% are monolingual Spanish (Spanish at home + limited English). In other words, of the 64% who speak Spanish at home, 95% also speak English well or very well. A cross-metro comparison adds a second finding: at the 2nd generation, every major Hispanic metro sits in a tight 80-94% Spanish-at-home band — Miami is not unusual in that regard. At the 3rd+ generation, the metros fan out: most lose 50-62pp from 2nd to 3rd+, while Miami loses only 25pp.
(2nd gen → 3rd+ gen)
(2nd gen → 3rd+ gen)
(2nd gen → 3rd+ gen)
share also fluent in English
How I label generations (used throughout the deck)
1st generation = born outside the US (or in Puerto Rico). 2nd generation = born in the US, with at least one parent born outside it. 3rd+ generation = born in the US, both parents also born in the US.
How I identify generation
The deck uses two rule variants depending on the question. The strict rule (Slides 3, 5, 14, 12, 9) requires both the household head AND their spouse to be US-state-born (BPL < 100) and applies to age 20-35; this is the version Miami's headline 64% 3rd+-gen Spanish-at-home rate comes from. Puerto Rico-born respondents are excluded from "native" for the language metrics (PR is itself a Spanish-speaking jurisdiction, so counting PR-born grandparents as US-native would conflate the construct); they ARE counted as native for the deck's structural metrics on Slides 14/8/12/9. The looser rule on Panel B only requires the head to be native, applies to all adults age 18+, and produces a higher 3rd+-gen Spanish-at-home rate (79%) on a wider universe (n=8,415 vs n=535 under the strict rule). The same strict rule applied to the 2nd generation requires the respondent to be US-state-born with at least one foreign-born parent — Panels E and F use this for the cross-metro 2nd-gen comparison.
What "speaks Spanish at home" measures
The ACS asks whether the person speaks a non-English language at home, what language, and how well they speak English. These are three separate questions. A person can speak Spanish at home and speak English very well — and most US-born Hispanic Spanish-speakers in Miami do.
Source
IPUMS USA: 1980, 1990, 2000 5% decennial census + 2010, 2020, 2024 American Community Survey 5-year samples (Panel A draws on the full 1980-2024 series; Panels B, C, D are 2024 ACS only).
Family structure predicts Spanish use; income and education don't. Continuing immigration is part of why Miami's level is high.
Within Miami, the most predictive thing to ask about an individual Hispanic adult to find out if they speak Spanish at home is who they married, what generation they are, what national origin, and how they self-report race. Across metros, Miami's high 3rd+-gen retention level is roughly what its high rate of continuing Hispanic in-migration would predict — the two move together across the 9 non-border metros (Pearson r = 0.86, 95% CI [0.46, 0.97]), though with only 8 metros once Miami is removed the relationship is suggestive rather than statistically confirmed.
What "predicts" means here (left chart)
For each candidate predictor, I computed the Spanish-at-home rate within each level of the predictor (PERWT-weighted, n ≥ 30 per cell), then took the gap between the highest-rate and lowest-rate levels. Self-reported English ability is excluded because it tautologically predicts language at home.
The interesting non-correlates
None of the standard structural-attainment indicators move the rate:
- Education: <HS 93.3% · BA 93.0% · Grad 91.6%
- Wage band: $1-30K 92.4% · $30-60K 93.8% · $100K+ 91.1%
- Homeownership: Owner 92.5% · Renter 93.5%
- Years in US (FB only): 0-5y 95.1% · 11-20y 95.6% · 31+y 94.8%
- Hialeah cluster: Hialeah 95.2% · Rest of MD 92.3%
How the replenishment test works (right chart)
X = share of Hispanic adults 18+ who arrived in the US in the last 10 years (in-migration intensity). Y = 3rd+-generation Spanish-at-home rate (same strict 2-parent-native rule used by Slide 3 and Slide 4, applied to the same county-level PUMA universes). Across the nine non-border metros, X and Y are strongly correlated (Pearson r = 0.86, 95% CI ≈ [0.46, 0.97] via Fisher transform). With Miami removed from the non-border subset, r falls to 0.58 across the remaining eight metros — and at n=8 that point estimate is not statistically distinguishable from zero (95% CI ≈ [−0.21, 0.91]). Border metros (McAllen, El Paso) sustain high retention via a separate mechanism — cross-border family and economic ties, not replenishment — and break the linear relationship if included (all-11-metro r = 0.45, 95% CI ≈ [−0.20, 0.83]).
Among the major Hispanic-population metros, Miami leads on education — and that lead is partly about who lives there.
Among the 11 major Hispanic-receiving metros (Miami, NYC, LA, Chicago, Houston, Dallas, Phoenix, San Antonio, Riverside-SB, El Paso, McAllen), Miami-Dade ranks first for college-educated US-born Hispanics. But comparing apples to apples on this single measure — Cubans to Cubans, Mexicans to Mexicans, etc. — most groups have higher college rates in other cities. The metro-level college lead reflects Miami's distinctive Cuban-heavy population, not Miami-specific within-group advantages on college specifically. (On the broader six-measure structural composite, the within-group picture flips back: see Slide 10, where Miami's Mexican-Americans rank 2nd of 11. Slide 9 shows Miami at 2nd of 19 on the all-Hispanic composite.)
Other measures Miami leads on (within the 11-metro panel)
Among US-born Hispanics across the 11 major Hispanic metros: labor force participation (87% vs 78-85% in the others), median wages ($63,585 vs $47,000-$62,000), management/professional jobs (52% vs 32-45%), lowest poverty rate, and highest non-Cuban Hispanic naturalization rate (64% vs 30-56%).
Where Miami doesn't lead even in the 11-metro panel
Homeownership: Miami's US-born Hispanic rate (60%) ranks third in the 11-metro panel, behind McAllen (65%) and El Paso (63%). Miami is the only metro in the panel where the US-born Hispanic homeownership rate exceeds the local non-Hispanic white rate, though.
Education and Spanish use are independent in Miami
It's tempting to read high Spanish use as a sign of low educational attainment, or high educational attainment as a sign of English-shift. Slide 7's correlate sweep shows neither holds in Miami. Among Miami Hispanic adults 18+, the Spanish-at-home rate is 93.3% for those without a high school diploma, 93.0% for those with a bachelor's, and 91.6% for those with a graduate degree — a 1.8-point spread. Education and Spanish use vary independently of each other.
Source
2024 ACS 5-year sample via IPUMS USA. The composition decomposition reweights each metro to a fixed pan-metro Hispanic-origin mix, isolating the within-origin contribution to metro-level gaps from the between-origin contribution driven by population mix.
If we combine non-language related measures of assimilation to a composite score, Miami performs strongly.
There are other signs that outside of language, Miami is doing well on assimilation measures. If we make a composite of six structural measures of assimilation (college degree attainment, labor force participation, median wage, management/professional employment, homeownership, and non-Cuban Hispanic naturalization) across the broader 19-metro panel (Miami's 11 major Hispanic-receiving peers plus eight metros where Hispanics are a smaller minority), Miami ranks second of nineteen, behind Atlanta and ahead of Washington DC. Whatever Miami's distinctive Spanish-language pattern represents, it sits alongside genuine structural strength on every other measured dimension.
How the composite is constructed
Six measures are computed for US-born Hispanic adults in each metro: BA+ rate (age 25+), labor force participation (25-64), median wage (employed 25-64, 2024 USD), management/professional share among employed, homeownership (heads of household), and non-Cuban Hispanic naturalization (foreign-born Hispanic adults excluding Cuban-origin). For each measure, metros are ranked best (percentile 100) to worst (percentile 0). The composite is the unweighted average of the six percentile ranks. Marriage and racial-identification measures are excluded for the purposes of this score, but comparative data can be found in the appendix. Equal weighting of these six fields is itself a value-laden assumption (it says every dimension matters equally to "assimilation"), but a weighting-sensitivity check confirmed Miami's top-3 ranking holds under several alternative weight vectors (e.g., 2x on wages-and-employment, 2x on homeownership, or up-weighting any single dimension by 50%).
About the 19-metro panel (and why it's not the 11-metro panel on Slide 14)
Eleven major Hispanic-receiving metros (the "headline" set used elsewhere in the deck) plus eight metros where Hispanics are a smaller minority: Washington DC, Boston, San Francisco Bay, Atlanta, Tampa, Las Vegas, Salt Lake City, San Diego. The expanded set tests whether Miami's structural strength survives a broader cross-section. It mostly does — Atlanta and DC outrank Miami because their small Hispanic populations are professionally selected (federal-government / corporate-professional jobs), but Miami still sits well above the next tier. Note on the metro-panel size difference vs Slide 14: Slide 14's 19-metro table can include any metro where at least one column's cell exceeds n=30. The composite here requires all six structural-metric cells to exceed n=30 simultaneously, which makes the panel slightly more restrictive — but in practice all 19 metros qualify on all six measures in the 2024 ACS sample.
The composite result holds when you compare like to like on national origin.
To isolate metro effects from origin-mix effects, I restricted the same six-measure composite to a single national-origin group — Mexican-Americans, the most widely distributed Hispanic origin across US metros. Miami still scores at the top of the panel. (Note the contrast with Slide 8: on the college-degree rate alone, Miami's Mexican-Americans don't lead. On the broader six-measure composite — averaging college, labor-force participation, wages, management/professional jobs, homeownership, and naturalization — they do. So Miami's structural strength holds within a single national-origin group when you look across measures, not just at college.)
Why this check
The composite on the previous slide could be doing two things: (a) genuinely measuring Miami's distinctiveness on structural outcomes, or (b) reflecting Miami's distinctive Cuban-heavy population mix. The Mexican-only cut isolates (a) by holding origin constant. Mexican-Americans are the most widely distributed Hispanic-origin group, so the panel of comparison metros is comparable across cells.
What the result shows
Chicago's Mexican-Americans score highest (73); Miami's Mexican-Americans are a close second (72) — within rounding of Chicago. The Mexican-American spread between El Paso (60) and LA (48) is wider than the Chicago-Miami gap at the top. Within a single national-origin group, the metro matters far more than how Hispanic the metro is overall, and Miami's Mexican-Americans land alongside Mexican-Americans in the strongest other metros.
Despite slow progress on language, many other structural measures of assimilation show promise.
As these slides show, Miami's unusual Spanish-language retention hasn't slowed its progress on other measures of assimilation. Native-born Miami Hispanics have seen durable gains in wages, education, and occupational status.
What's measured
Five structural outcomes for native-born Hispanic adults (college degree, labor force participation, management/professional employment, homeownership, mean wage), each computed for US-born Hispanic adults age 25-64 in Miami-Dade across six census/ACS samples (1980, 1990, 2000, 2010, 2020, 2024). The bottom panel additionally shows the college-degree rate broken out by generation among young adults age 20-35.
How "Miami-Dade" is defined across decades
The boundaries of the "Miami metro area" expanded in the early 2000s, which complicates a 44-year comparison. For 1980-2010 I use the IPUMS metro-area code (which lined up with the county before the metro expanded). For 2020 and 2024 I reconstruct the county from smaller geographic units (PUMAs). The two methods overlap for 2010 and produce consistent numbers, so the time series isn't broken by the geography change.
How generations are identified
For the generational panel only: I identify 2nd-generation and 3rd+-generation Miami Hispanic young adults (age 20-35) by finding them in households where they live with a parent. Where the young adult was born in the US and at least one parent was born outside it, I call them 2nd-generation. Where both parents were also born in the US, I call them 3rd+-generation. (This methodology is described in detail on Slide 6.)
Source
IPUMS USA: 1980, 1990, 2000 5% decennial census + 2010, 2020, 2024 ACS 5-year samples. Wages adjusted to constant 2024 dollars using the standard CPI-U inflation series.
Miami Hispanics have converged with national non-Hispanic whites on most structural measures.
Miami Hispanics aren't just doing better than Hispanics in other metros. On four of five measured dimensions Miami Hispanics have caught up to (and in a few cases, slightly overtaken) national non-Hispanic whites since 1980. The lone exception is mean wages, where the absolute gap has widened modestly even though both populations gained substantially in real terms.
Universes
Miami Hispanic: US-born Hispanic adults age 25-64 living in Miami-Dade County (1980-2010 IPUMS METAREA, 2020/2024 PUMA reconstruction). National non-Hispanic white: NHW adults age 25-64 nationwide, non-group-quarters. Both populations measured the same way in each sample.
Measure definitions
BA+ (EDUCD ≥ 101); labor force participation (LABFORCE = 2); mean wage (PERWT-weighted INCWAGE among employed, CPI-U-adjusted to 2024 dollars); managerial/professional (OCC1990 003-200 among employed); homeownership (OWNERSHP = 1 among heads of household, RELATE = 1).
Why this comparison
Slide 11 shows Miami Hispanics rising over 44 years; Slide 9 shows them ranking 2nd of 19 metros on the cross-sectional composite. Neither answers the natural follow-up: relative to whom? Pairing against the national non-Hispanic white reference population is the most-cited benchmark in the assimilation literature. The answer here: Miami Hispanics have converged with NHW on four of five measures, with the wage gap as the standing exception.
Caveats
The Miami population is metro-specific and the NHW population is national; some convergence reflects city-vs-national differences (urban service economies have higher LFPR, higher cost of living, etc.) rather than Miami-Hispanic-specific gains. The wage gap is partly a national whites-pulling-away-faster story (the same gap widened modestly between national Hispanics and national NHW over this period); not unique to Miami.
Compare any two groups directly.
This interactive tool lets you pick any (Hispanic origin × metro) pair for Group A and any (Hispanic origin × metro) pair for Group B, then compare them across 15 measures spanning education, jobs, marriage, racial identification, language, and military service — all from the 2024 ACS 5-year sample. An optional toggle adjusts the non-Hispanic-white intermarriage rate by the local NHW share (the propensity index from Slide 9). Cells based on fewer than 30 sampled people are marked too small to report.
| BA+ (age 25+) | 45.2% |
| Mean wage | $79.8K |
| Homeownership | 67.2% |
| Marries within own origin | 65.0% |
| Pan-Hispanic marriage | 22.0% |
| BA+ (age 25+) | 18.8% |
| Mean wage | $59.1K |
| Homeownership | 55.1% |
| Marries within own origin | 72.0% |
| Pan-Hispanic marriage | 7.7% |
| BA+ | +26.4 pp |
| Mean wage | +$20.7K |
| Homeownership | +12.1 pp |
| Marries within own origin | −7.0 pp |
| Pan-Hispanic marriage | +14.3 pp |
- Mean wage and BA+ rate among Cuban-Americans in Miami vs Mexican-Americans in Houston
- Pan-Hispanic vs same-origin marriage rates for Dominican-Americans in NYC vs Miami
- Spanish-at-home rate among Cuban-Americans in Miami vs Salvadoran-Americans in LA
- Non-Hispanic-white intermarriage rates with and without the local-availability adjustment toggle
- Veteran rate or "white-only" race answer for any (origin × metro) pair
Five dimensions of assimilation, summarized across 19 metros.
This chart also shows how Miami compares on measures of pan-ethnic outmarriage and racial identification.
What's measured
For each of 19 metros, one representative number per dimension. US-born Hispanic adults only. Numbers come from the 2024 American Community Survey 5-year sample.
About the metro set
Eleven major Hispanic-population centers plus eight additional metros where Hispanics are a smaller minority (Washington DC, Boston, San Francisco Bay, Atlanta, Tampa, Las Vegas, Salt Lake City, San Diego). Residential dissimilarity is computed at the PUMA level for all 19. PUMAs contain ~100,000 people each, so PUMA-level dissimilarity systematically understates within-metro segregation; the cross-metro comparison is more reliable than the absolute values.
Sensitivity caveat on the military-service column
Miami's rank on military service (veteran rate among US-born Hispanic adults) is partly driven by whether Puerto Rico-born respondents are classified as native. The table treats PR-born as native (consistent with this slide's structural-metric pipelines, where the Spanish/English distinction isn't germane — note that the deck's strict 3rd+-gen LANGUAGE rule on Slides 3 / 4 takes the opposite choice, excluding PR-born, because counting Spanish-speaking PR as US-native would bias the language test). Under the alternative spec that excludes PR-born from the native universe here, Miami's veteran-rate rank drops further (roughly 8th → 10th of 11 in the 11-metro sub-panel). The deck's "sits below average on military service" framing holds under both specifications.
Source
2024 ACS 5-year sample via IPUMS USA. Each measure is computed on the universe appropriate to it (e.g., labor-force participation on age 25-64; marriage outcomes on married adults; homeownership on heads-of-household).
Who marries whom: marriage patterns among US-born Miami Hispanics.
Miami's US-born Hispanics marry across national-origin lines (Cuban-Colombian, Dominican-Mexican, etc.) more than in any other metro: 33% of marriages are pan-Hispanic, vs 3-20% elsewhere. Marriage to non-Hispanic whites is comparatively low in absolute terms (11%), but high once you adjust for the demographics of the local marriage market.
What's measured
For married US-born Hispanic adults, the race/ethnicity of their spouse. Four categories: same Hispanic origin (endogamy), different Hispanic origin (pan-ethnic), non-Hispanic white (mainstream), non-Hispanic non-white (other).
Why mainstream marriage rate is tricky
The local pool matters. NYC and Phoenix have larger non-Hispanic white populations, so mainstream-white marriage is mechanically more available. Miami-Dade is roughly 20% non-Hispanic white; that limits the available marriage pool. In raw terms, Miami ranks 9th of 11 on mainstream-NHW marriage rate — but once you adjust for the local-NHW-share availability (the propensity index reported on Slide 9), Miami ranks at the top of the panel. The deck uses the propensity index as the more meaningful comparison; the raw 9-of-11 ranking is disclosed here for completeness.
One thing to know
Pan-Hispanic marriage means a Cuban marrying a Dominican, or a Mexican marrying a Puerto Rican — distinct national-origin groups within the broader Hispanic category. It's only possible at scale where multiple Hispanic origin groups co-locate, which Miami's population mix uniquely supports.
How marriage loops back into language
The correlate sweep on Slide 7 found that spouse identity is the single strongest individual-level predictor of Spanish-at-home use among Miami Hispanic adults. Hispanics married to a Hispanic foreign-born spouse use Spanish at home 96% of the time; Hispanic native-born spouse, 90%; non-Hispanic spouse, 74%. The 22-point gap between Hispanic-foreign-born and non-Hispanic spouses exceeds the spread across any age band, income bracket, or neighborhood. Marriage isn't just an assimilation outcome on its own — it shapes the household language environment that Slides 6 and 7 measure.
Source
2024 ACS 5-year sample via IPUMS USA. Spouses are linked within each household by reading both partners' census records together. Intermarriage shares are computed over married adults aged 25-64.
Universe definition, weighting basis, and suppression rule for every percentage cited in the deck.
Every percentage in the deck has a precise universe (who is counted), a sample year, a PERWT-weighting basis, and a suppression rule. The methodology appendix is a single reference table covering every headline metric — organized slide by slide, with one row per number cited.