New Evidence of Generational Progress for Mexican Americans

U.S.-born Mexican Americans suffer a large schooling deficit relative to other Americans, and standard data sources suggest that this deficit does not shrink between the 2nd and later generations. Standard data sources lack information on grandparents’ countries of birth, however, which creates potentially serious issues for tracking the progress of later-generation Mexican Americans. Exploiting unique NLSY97 data that address these measurement issues, we find substantial educational progress between the 2nd and 3rd generations for a recent cohort of Mexican Americans. Such progress is obscured when we instead mimic the limitations inherent in standard data sources.

That is by Brian Duncan, Jeffrey Grogger, Ana Sofia Leon, and Stephen J. Trejo in a recent NBER working paper.


Here come the haters: 'Methodology wrong.....sociology not a science...can't happen genes don't change in 2 generations...yadda yadda'

Kind of like how you shriek Hazel Meade-like at any studies casting doubting on global warming. but that's different I guess.

Poor Sam. He can't imagine how that looks to a random normal person who happens by, and gives a shot at reading the comments.

And of course it goes downhill from there.

Ignorant, lonely, witless Nathan.

Bzzzt. Try again.

The fact that you can believe random b.s. of your own invention should be a warning to you .. but you just don't have that feedback circuit, do you?

Wrong target for your spluttering, Sam. I'm a Lomborg/Ridley warming realist.

Based on the comments so far, too optimistic.

revealingly for leftists optimism is generated by a failure to deal with statiscal realities. (Hint it's why you keep getting stomped on politically. Please keep it up I'm really really not tired of winning yet.)

The 2018 midterms are going to make you sad.

The Sailerites could always just trot out the line that this matches up with findings from ethnic attrition, where white Mexican-American stop identifying as Mexican-Americans by the 3rd generation or so as their family intermarries with non-Hispanic whites. So this study would just capture the fact that white people with Mexican ancestors tend to be like white people without Mexican ancestors.

Sure, it's called the melting pot. Nice to see proof it's still a thing.

Be careful you don't get hoisted by your petard there; isn't the melting pot ideal now considered racist? Though it certainly will be funny, viewed from a certain vantage point, if it is the descendants of Hispanic (and Asian) immigrants that end up preserving the US's status as a white majority nation. That would certainly piss off large elements of both the right and the left, at the very least.

I was talking to a high school teacher at a mostly Hispanic school who said that in 2005-2006 during the Housing Bubble, his better male students tended to drop out to get construction jobs leaving only the lazier boys to finish school. But when the jobs disappeared after the bursting of the Bubble, then everybody hung around to graduate and go to junior college.

A teenage boy would learn more working in construction than he would in high school. He would become a skilled worker, rather than an unskilled one. Of course, he wouldn't learn the important skills needed to become a desk-bound academic or bureaucrat, the goal of his Anglo classmates.

The average second generation Hispanic is reading at a fifth grade level when he enters ninth grade. Call me a bleeding heart leftist but I really don't think that's good enough.

It's a pity they're stuck in schools run by the NEA then. Rewarding failed teaching shouldn't be the solution.

The "average" second generation Hispanic? A meaningless assertion. Why do they, or anyone else for that matter, need to be able to read? They have access to television, after all. Sean Hannity and Joy Behar can tell them all they need to know. The percentage of Hispanic immigrants with home delivery of the NYT or WaPo is probably below that of intelligent real Americans because of their low reading level but then they're not being influenced by elite propaganda, which apparently is the real problem.

I call B.S. on the fifth grade reading level

Artimus said, "I call B.S. on the fifth grade reading level".
If you mean, "I call B.S. on average ninth grade Hispanics having a fifth grade reading level." you would be wrong.
It is also known that the average Hispanic high school graduate's math and reading score is at the average White's middle school level.

I just spend five minutes Googling and the answer becomes apparent:

Here is an NPR link showing how a Washington, D.C high school graduation and by implication, future community college graduation has become a farce.

I have always been a bit confused about this "reading at X grade level" thing. You can only really get so good at reading. I don't think my ability to read changed from the age of 6 on. If you're still learning to read in 9th grade don't you have a serious problem?

It's a proxy for intelligence. Basically asking how sophisticated a written argument a person can understand.

"Standard data sources lack information on grandparents’ countries of birth, however, which creates potentially serious issues for tracking the progress of later-generation Mexican Americans."

To translate that means after they enjoy gains in education and resources they say they are white.

When Lee Trevino won the United States Open golf tournament in 1968, he told reporters he's going to get so rich he'd stop being a Mexican and start being a Spaniard.

Table 6 in the paper reports average years of schooling by number of Mexican-born grandparents for third generation Mexican-Americans (i.e. those whose parents were born in the US):

1 Mexican grandparent: 13.81 years

2 Mexican grandparents: 13.63 years

3 Mexican grandparents: 12.35 years

4 Mexican grandparents: 11.81 years

(small samples though)

Mexicans who out-marry tend to be higher-achievers.

Like when I was a kid in the 1960s, my Swedish-American cousin was married to a Chicano guy who had a good white collar sales job with Kraft and they owned a nice house with a pool. Their Swedish-looking kids identified as regular Americans and their Mexican background was mostly spicy food.

There's a real sense that Mexican Family Values tend to hold kids back. The Pixar movie "Coco" insinuates how suffocating Mexican familism can be to a talented kid.

In California, intermarriage rates for Hispanics have been dropping in this century, mostly because there are just so many other Hispanics around to marry.

I wonder how common Mexican-Filipino marriages have become.

One of the themes of the musical sitcom "Crazy Ex-Girlfriend" was that the unambitious, potentially downwardly mobile Jewish girl in West Covina, CA was likely to get sucked into the Filipino-Mexican lower middle class of the SoCal suburbs if she didn't dump her cheerful but clueless Filipino boyfriend and marry either the alienated rich kid from Beverly Hills who gets her or the sexy USC WASP bro.

I thought filipinos did really well income wise in the U.S., like above white averages.

"Substantial progress" is very relative. 4+ generation Mexicans have a similar level of educational attainment as African Americans, e.g. 20% graduate from college versus 40% of whites.

...creates potentially serious issues for tracking the progress of later-generation Mexican Americans.

Gee, that's too bad. Tracking things, especially progress, whatever that means, is so very important.

What does substantial progress mean? Improvements to what level? Without numbers you can word things to fit whatever narrative you want.

The paper is available at Tyler's link. See the tables at the end of the paper. Mexican-Americans are like African-Americans, a permanently poorly performing minority.

Hush you are making the resident hard leftists all pessimistic.

The tables say that, measured as actual people, they converge, but the more successful ones stop calling themselves Mexican-American and in some cases stop using the Hispanic ethnic qualifier, only race.

This is similar to, but an opposite effect, of how unsuccessful European-Americans with low levels of schooling tend to shift their answers to ancestry questions on the Census from English, Scotch-Irish, French, German, or what have you to just "American" or "White" (on the ancestry question, not race.) That is part of why specific pride in being white instead of a more specific ethnic background is correlated with low IQ and low levels of schooling.

Or, rather, the successful ones intermarry with non-Hispanic whites to the extent that it makes no sense for them, genealogically, to call themselves Mexican-American.

Does anybody have any idea how many Mexicans there would be in the U.S. if there was a one drop rule like with blacks? There sure are a lot of self-identified Mexicans (over 36 million according to Pew) even without a one-drop rule.

How many would there be from a genealogical perspective (i.e., counting half-Mexicans, as half, etc.).

Interesting point, John.

I didn't know because it said the paper had a purchase cost.

I'm pretty sure it was freely available when Tyler published his post. Maybe NBER slapped a price tag on it because of traffic from MR? Use Sci-hub or just Google to read the paper.

I wonder if this is one of the reason why there was such an Illegal Immigration backlash in 2016 with Donald Trump. When Latino-Americans starting catching up with white populations, some white populations did not like that.


As pointed out above the only people that Mexican immigrants have caught up with are black people- and they didn't vote for Trump. Troll harder.

Good that they are making progress. I have 2 'nephews' whose Mom is from Mexico. They are in Texas. School system there puts anyone who can speak Spanish into a hybrid system of classes taught in both languages. It's cool that they are very fluent in Spanish and always will be, but I can't help to think that their education should be English only as soon as they can handle that (they can, but for other students).

Being multi-lingual is a recipe for schizophrenia as can be seen by taking a casual look at Europeans.

>Nearly, three-quarters (72%) of married, U.S.-born Asians, and over half (52%) of U.S.-born Latinos are interracially married, and most often, the intermarriage is with a white partner.

Louis CK is a Latino by some definitions but would anyone really identify him as such? Would he self identify as Latino?

High rates of intermarriage between Mexican Americans and White Non-Hispanics are rapidly blurring the lines between "Latino" and "White". It is NOT like black-white marriages in which the child almost always is identified as "black" (e.g., Barack Obama).

If you genuinely think of yourself as a 4th generation Mexican American with 100% Mexican heritage, it's probably because your family didn't integrate well into broader society.

Mexican Americans are marrying into white America at about the same rates as Italian Americans did.

Which is to say, what perpetuates "Hispanic" as a politically-relevant distinction is immigration. if we adopted, say, a strictly humanitarian immigration policy and passed over Mexicans in favor of low-caste Indians, soon enough Hispanics would be as politically-relevant as Italians.

If you genuinely think of yourself as a 4th generation Mexican American with 100% Mexican heritage, it’s probably because your family didn’t integrate well into broader society.

Blue ribbon for most preposterous statement of the day. A huge portion of what's now the US was once Mexico. The Mexicans that live in that area were and are the "broader society".

Blue ribbon for most preposterous statement of the day. A huge portion of what’s now the US was once Mexico. The Mexicans that live in that area were and are the “broader society”.

That 'huge portion' had a five digit population in 1848. The number of chicanos in the United States predominantly descended from old criollo, mestizo, and mission Indian families is just about nil.

Bunk. My wife has a dozen Mexican-American cousins and they range from engineers, to mechanics to teachers to accounting clerks and bank managers.

The folly of trying to describe the Hispanics as a single entity, and the surprises when they progressed. Previously I showed that in California with significant Hispanics population, the population dynamics for those national merit scholar semifinalists are such that there are two choas theory strange attractors at different performance levels that two distinctly different ability distribution groups will be resulted. The large sample of the ACT score distribution for the Hispanics with two distinct but slightly overlaping bell shape distributions confirms that.

It is clear from the chart that the 'race average score' is very close to the dip valley, a value which is neither here nor there. Inferences based on that are mostly meaningless. The separation of the two sub-populations is most probably due to assortative mating as the US Whites also shows similar bimodal distributions but the admixing might have began 300 years later than the Spanish Conquest, the elite subpopulation is at 56% rather than that for the Hispanics at 35%.

The bimodal distribution means that after 500 years there is still significant segregation of population and there are still significant relatively 'pure' European linages. The modal IQlike value for the Hispanic elite subpopulation is close to the IQ for Spain. The White performance distribution in another 300 years might be the same as that for Hispanics today.

The interesting part is that the dip valley seperating the bimodal distributios for the Hispanic, White+, Black and Asian Americans are all at ACT score 20, i.e. there are significant distinctly separated assortative mating at the elite and lesser levels and the self identified 'races' are more or less random dependent on the 'races' of the fathers or the single parents. Performance-wise the abilities of the lesser clusters of the different 'races' are very similar. So are those of the elite clusters.

Instead of the often quoted White-Hispanic IQ gap of ~1 SD, there is little difference between the modal IQlike values for both elite subpopulations (0.18 SD) or for both lesser subpopulations (0.07 SD). While at the same time the difference between the elite and lesser subpopulations for the same ethnics are respectively (1.29 SD) and (1.13 SD) for Whites and Hispanics. Though the practical differences are very small, the statistical ANOVA test still stubbornly refused to acknoledge them as similar within the lesser or elite clusters separately. This 44% lesser White+ cluster have below college readiness score of 22 and without the privileges of the affirmative actions and diversity actions compare with those similar Blacks and Hispanics might be the core supporters of Trump.

It is this sort of data that I start to accept that the self identified 'race' is a social construct where the between variances are less than the within subpopulations. It is interesting that it is this sort of data that can support the left's narratives are ironically suppressed by the left as the data at the same time shows that instead of just differences between the races there are even larger differences within the races. The above data were for years publicly available for direct download with simple clicks without any further procedures. However, recently after someone pointed out the availablity of the reports, they mysteriously disappeared.

In US self identified 'race' seems to be a social construct as shown by the data. There is really only class difference. However politics will skew that to the extreme.

Side note: At the elite cluster, both White+ and Hispanics have the same modal value (at the highest freq distribution) at ACT score 24. Note also that Blacks also have a small but noticeable presence at the very high right hand end. At the right hand tail end where the distribution thinning out, the Asian American cluster decides to have a 'third' bell shape cluster at IQlike modal value at 137.

To try and simplify the argument; among all self-identified "races" within the US, we see a bi-modal distribution of standardized test scores within that "race's" population. The medians for the "elite" subgroups and the "non-elite" subgroups are more or less the same for all of the races. Therefore... this is where I lose the thread of the argument. I will admit that if this is true, it is a striking phenomena, one which would be worth investigating further. But what exactly it implies isn't entirely clear. It could be some sort of shared elite and non-elite culture, more or less. It could be some sort of biological discontinuity.

The raw US personal income distribution has been known to be multimodal. If education attainment is part of the causes for the multimodality there should be some standard test distribution that show that. However, most testing authorities have guidelines to ensure that their results show unimodal bell curve. Thus I was looking for raw distribution data that show multimodal properties.

The extent that researchers have to go to smooth out the distributions is documented in this BIS (Bank of International Settlement, the central bank for the gloabal central banks) report, The author called this the Iterative surgery procedure.

The BIS report also noted that """However, the trend was reversed in the 1990s, the low-end spike re-emerged and a large amount of population mass moved from the middle and upper-middle relative income ranges to the lower-middle range. The observed low-quality convergence might not be desirable, with the middle class potentially disappearing into the lower range.""" So could the ACT distributions for the White+ and Hispanic composition percentages shows different stages of this movement and cause this change, together with other factors like the outsoucing of jobs to oversea and the emergent of lesser number of high tech jobs that require higher education attainment? It might be harder to explain this multimodalities from a unimodal bell curve.

So, again, let me see if I am following. Not only are standardized test scores multimodal, but so is the income distribution in the US. This suggests that there is some sort of interaction between the test scores and the income distribution. Perhaps the two reinforce each other somehow? Again, there would be so many steps in the causal chain to suss out that I am not sure how to interpret the raw data, but do find it interesting.

Comments for this post are closed