Disappearing Cajuns and Creoles? Ethnic Identity and the Limits of Census Data

Historians and math do not always go well together. In fact, historians are notoriously bad when it comes to math. There are exceptions, as in the case of those odd creatures, quantitative historians, who use a mathematical approach to understanding History. (See, for example, Fogel and Engerman’s noted quantitative history Time on the Cross: The Economics of American Negro Slavery, originally published in 1974.)(1)

“Historians and math
do not always go well together”

While I never excelled at math (an understatement), I nonetheless use quantitative analysis, albeit sparingly, in my books. I did so perhaps most notably in The Cajuns: Americanization of a People, which I sprinkled lightly with numerical data gleaned from U.S. censuses, including, but not limited to, the most recent one, the 1990 Census. (I wrote in the late 1990s, before the 2000 U.S. Census had been compiled.)(2)

My 2023 book The Cajuns:
Americanization of a People

Although only about a quarter-century ago, I wrote The Cajuns at a time when census data was not, unlike today, readily available on the Internet — at least not in any detail. Researchers instead had to visit a brick-and-mortar library and consult multi-volume censuses printed on actual paper as well as in other forms of media. Moreover, finding sought-after data could be extremely difficult because, as one library notes, “the 1990 census filled hundreds of volumes, CDs and numerous tape files. . . .”(3) Even then, the required data might not actually exist in any of those published sources.

In that case, there was an alternate source of census data: the Public Use Microdata Sample (PUMS), a digital database of raw census data based on a 1-percent, 3-percent, or (as I myself used) 5-percent sample of respondents in specific geographic regions. This is important, because while the short-form 1990 U.S. Census — the version sent to most American households — asked recipients how they identified racially (Black, White, Hispanic, and so on), it did not ask how they identified ethnically (Italian, Jamaican, Filipino, Dutch, Norwegian, or any number of other ethnicities). However, the long-form 1990 U.S. Census — received by 5 percent of U.S. households and gathering the actual PUMS data — did ask respondents to identify their ethnic ancestries. Using complicated formulas, the U.S. Census Bureau could then extrapolate from that sample to provide ancestral data about all U.S. households.

Figure 1: 1990 PUMS data, for Cajuns,
by Louisiana parishes and select Texas counties,
Acadiana parishes in bold,
(click to enlarge).
Image source: author's defunct website.

PUMS data, however, was hardly perfect. Besides the fact it could never provide more than an estimate (one would hope an accurate estimate), the general public often could not access PUMS data, which required somewhat powerful computers (for the time) loaded with the proper, and rather complicated, software. As such, PUMS data generally had to be accessed through an institution, such as a university. Moreover, it was not enough to access the PUMS database: rather, you also had to know, or know someone who knew, how to program the database to obtain the sought-after data. And because that output appeared in a form hardly describable as “WYSIWYG” (“what you see is what you get” or, more plainly, a self-explanatory format), you also had to know, or know someone who knew, how to interpret that very user-unfriendly data printout.

User-unfriendly PUMS data printout,
1990 U.S. Census

Fortunately, I had access to PUMS data through the Sociology Department of Texas A&M University, where I pursued my doctoral degree in History with a minor in Rural Sociology of Minorities (very useful if, like me, you are studying the Cajuns and Creoles of rural and small-town south Louisiana). It was my Sociology professor, Dr. Rogelio Saenz, who taught me about PUMS data and helped me to obtain and interpret the PUMS data I sought.

For example, I might have asked the PUMS database to report how many Louisianians in 1990 identified their ancestry as primarily Cajun or Acadian (the latter of which I interpreted to mean “Cajun”)* and of those how many spoke French as their first language in the home; or of those how many identified as World War II veterans; or even of those how many both spoke French in the home and identified as World War II veterans. To continue with this example, I might (as I did) then extrapolate backward and estimate the total number of Louisianians of Cajun ancestry who fought in World War II and how many of those “Cajun GIs” spoke French in wartime.(4) (A note: I did not compile 1990 PUMS data about Creoles, though that raw data did exist in the PUMS database. I chose not to do so because, while my dissertation topic originally focused on Cajuns and Creoles, I found it necessary to winnow my focus solely to Cajuns — otherwise I believed the scope of my dissertation would have been too unwieldly. A fragment of my early research focusing on both Cajuns and Creoles can be found here.)

Intriguingly, the 1990 PUMS data revealed that there were 432,549 persons in Louisiana who self-identified as primarily Cajun compared to 668,271 such persons in the entire U.S. (see Figure 1). Many of those out-of-state Cajuns no doubt emigrated from Louisiana in the wake of the recent “oil glut” that shattered the state’s oil-dependent economy in the mid- to late 1980s.(5)

This PUMS data also identified a number of quirks among the assumed Cajun population. For example, it revealed that 10 percent of all Louisianians considered themselves as primarily Cajun. And while as mentioned 432,549 persons in Louisiana listed their primary ancestry as Cajun, another 25,000 listed their secondary ancestry as Cajun. Vermilion Parish, in south-central Louisiana, possessed the largest percentage of persons identifying primarily as Cajun, about 50 percent. Also, the census suggested that most Cajun respondents had not strayed far from their ancestral south Louisiana homeland: of the 668,271 persons throughout the U.S. identifying as primarily Cajun, 77 percent resided in Louisiana or neighboring Texas.(6)

Moreover, few persons in New Orleans considered themselves Cajun — despite the media and tourism industries painting New Orleans as a “Cajun” city, a claim that appeared ad nauseum during the Cajun fad of the 1980s. Yet in 1990 only one percent of respondents living in New Orleans proper (Orleans Parish) marked “Cajun” as their primary heritage. The percentage remained low (7 percent or less) in parishes surrounding New Orleans. Indeed, census data showed that New Orleans boasted about the same concentration of Cajuns as Houston. As I wrote in my book The Cajuns, “Houston, Texas, possessed nearly the same percentage of Cajuns as New Orleans — less than a quarter of a percent difference — and on a per capita basis Houston actually boasted 4.5 times as many Cajuns.” And yet, unlike New Orleans, the media and the tourism industries did not portray Houston as “the center of the Cajun universe.”(7)

❧

Over the past twenty-five years since I did my PUMS research, the Internet has grown exponentially, along with the amount of useful (and not so useful) data it offers to scholars and laypersons alike. In fact, the U.S. Census Bureau’s website now affords the general public access to extremely detailed census results in digital format — meaning there is arguably no longer a need to navigate countless hardcopy volumes to locate sought-after data. With a few clicks of a mouse, that information can now be accessed 24/7 using any computer (as well as digital tablets and cell phones) linked to the Internet.

The task, however, still requires a degree of skill and patience because the required data might not at once be discernable. As with the old PUMS database, researchers must sometimes configure the census.gov interface to reveal the desired information — and just how to do that is not necessarily self-evident. In addition, it’s easier than ever to become mired in the sheer volume of census data now available to everyone. (Fortunately, I have found that Census Bureau experts promptly answer requests for help submitted by email.)

This being said, it is interesting to compare current U.S. census stats for Cajuns with the 1990 census stats I consulted a quarter century ago. Moreover, it’s possible to conjure up similar stats for the other group of interest to me, Creoles — by which I mean, unless otherwise stated, Creoles by any definition, and regardless of color. (In this essay I won’t parse the word “Creole,” much less the assertion, which I embrace, that “Cajuns are a type of Creole”; but my essays on these topics can be found here.)

Besides conducting the usual decennial census, the U.S. Census Bureau now compiles something called the American Community Survey (ACS). The Bureau describes the ACS as “an ongoing survey that provides vital information on a yearly basis about our nation and its people.” (You can read about the differences between the decennial U.S. Census and the ACS here.)(8) And the ACS does count the number of respondents who identify as “Creole,” as well as those who identify as “Cajun.” Accessing this sought-after data, however, is somewhat complicated: while population stats for Cajuns are fairly easy to find, those for Creoles are, unfortunately, buried a little deeper in the raw data.

According to one specific ACS table available on the U.S. Census Bureau website, namely, Table B04006, titled “People Reporting Ancestry” — accessible here, or see Figure 2 — 107,553 persons throughout the U.S. identified their primary ethnicity in 2020 as “Cajun” (give or take about 4,222 persons according to the margin of error — but I won’t list margins of error from here on out; you can find them on the original charts). Of these, 55,022 lived in Louisiana, or 51.15 percent of all Cajuns; while 19,141 lived in neighboring Texas, or 17.79 percent of all Cajuns. (Note I am looking only at responses for primary ethnicity, not secondary, which I leave for others to explore.)(9)

Figure 2: Table B04006,
“People Reporting Ancestry,”
from the 2020 American Community Survey,
Cajun stats highlighted
(click to enlarge).

Already something odd is discernible: only 55,022 Cajuns exist in the entire state of Louisiana? And those Cajuns comprise only 51 percent of the total number of Cajun people in the entire U.S.? These stats seem far too low. But more on this shortly.

What we don’t find on Table B04006, however, is any reference to persons who identified as “Creole.” It seems the Census Bureau relegated those respondents to a catch-all “Other” category. As a result, Creoles went uncounted on this table as a standalone ethnic group.

This does not mean the U.S. Census Bureau overlooked Creoles in 2020. In fact, self-described Creoles were counted, albeit on a different ACS table for 2020: namely, a custom-generated table compiled using the Microdata Access Tool (MDAT), which drew its output from the ACS 5-Year Estimates PUMS dataset. The results can be accessed here, or see Figure 3.(10)

According to this MDAT table, 7,887 persons in Louisiana identified their primary ethnicity in 2020 as “Creole,” while, for comparison, 6,308 persons in Texas likewise identified their primary ethnicity as “Creole.” This reflects the emigration of south Louisiana Creoles to the Lone Star State, mainly its southeast region, over many generations going back well into the early to mid-20th century.

Again, however, these stats for Creoles, like those for Cajuns, seem incredibly low.

Figure 3: MDAT table for Creoles,
Louisiana and Texas, 2020,
Creole stats highlighted
(click to enlarge).

Other conundrums also present themselves. Why, for example, are there so many fewer Creoles than Cajuns? If anything, I would expect more Creoles than Cajuns, not the other way around.

In my opinion these issues reveal the limitations of self-reported ethnic and racial data. A major issue, for example, is that a person of Creole ancestry might correctly identify as “Creole,” but just as correctly identify as “African American” or “Afro American” or “Black” or “Afro” or “Negro” (all labels represented on the MDAT table). They might also identify as members of those annoyingly generic, ill-defined, and I would assert almost meaningless “ethnicities,” namely “American” and “North American.” They might also correctly identify as “Cajun” or “Acadian” or “French” or any number of other ethnicities they might embrace. And this is considering only Creoles of African descent, not Creoles who identify as White and might correctly claim a variety of ancestries in their own right — thus further complicating the task of ethnic self-identification.

The issue is the same for Cajuns. They might correctly identify as primarily “Cajun” or “Acadian” or “French” or “Canadian” or “French Canadian” or “Creole” (again, Cajuns can be viewed as a type of Creole) or “American” or “North American,” and so on.

This could explain why, whether we compare the 1990 census results to those on the 2020 ACS Table, or the 2020 ACS table to the MDAT table for that same year, we end up with notable data discrepancies. 

For example, the 2020 ACS table counted 55,022 self-described Cajuns in Louisiana — but what happened to the other 377,527 who identified as such in 1990? Similarly, according to the MDAT table there were 45,884 self-identified Cajuns in Louisiana in 2020 . . . about 9,000 less than the 55,022 tallied for the same year on the separate ACS table (see Figure 4).(11)

Why the difference between these two 2020 sources, both issued by the U.S. Census Bureau?

Figure 4: MDAT table for Cajuns,
Louisiana and Texas, 2020,
Cajun stats highlighted
(click to enlarge).

As a demographic statistician at the Bureau kindly explained to me, “For relatively small groups there are sometimes differences in results depending on the data set used. The microdata tool uses the American Community Survey (ACS) PUMS while Table B04006 uses the full ACS data set. This may be part of the reason why there would be a difference in a relatively small group such as Cajun. Another reason is that 2020 was a difficult year for data collection due to the [COVID] pandemic. The data from 2016-2020 for smaller groups may be less reliable than normal.”(12)

This is, however, not the first time incongruities have been noticed by those using censuses to study ethnicity in south Louisiana. As I wrote in my Americanization book:

[R]esearchers have discovered a major discrepancy between the 1990 census and preliminary results from the 2000 census. The 1990 census counted over 400,000 Cajuns in Louisiana, while the 2000 census counted only about 40,000 — a roughly 90 percent decline in only ten years! The U.S. Census Bureau clearly miscounted, either in 1990 or 2000 (or both), for the disappearance of almost the entire Cajun population in only a decade is highly improbable. . . . Louisiana historian Carl A. Brasseaux . . . discounted the 2000 statistics, noting wryly that there are probably 40,000 Cajuns on the north side of Lafayette Parish alone. Non-academics also have scoffed at the 2000 statistics. Lafayette’s Daily Advertiser [newspaper] ridiculed the figures as “cockeyed” and observed “Our government advises [us] that there aren’t as many Cajuns . . . as we saw dancing in the streets during festival time.” “If You’re One of 365,000 Missing Cajuns,” ran one of its headlines, “Please Send up a Flare.” Asked another newspaper, “Where Did All the Cajuns Go?”(13)

I in no way criticize the U.S. Census Bureau. In fact, I would now assert that such discrepancies in census data stem in part — perhaps in large part — from the fact that we are dealing (needless to say) with human beings; and human beings are innately subjective creatures. As such, they are not only difficult to pin down, but very much dislike being pinned down. As shown, the very multiplicity of correct answers a Cajun or Creole might give to an ancestry query is also problematic. Likewise, it is possible long-form census recipients chose, say, the first ethnic label that came to mind, rather than the ethnic label they themselves might have genuinely regarded as most suitable or accurate. Moreover, some, perhaps even many recipients simply might not have answered the ancestry question, for whatever reason (confusion, impatience, privacy concerns, etc.).

Census data can be useful for indicating general demographic trends, but it can prove misleading if always taken at face value. (Again, only 7,887 Creoles in all of Louisiana in 2020? And only 55,022 Cajuns?) This, however, is where narrative history can complement census data, as well as complement more quantitative approaches to history in general. By drawing on traditional and — as I like to do — not-so-traditional sources, history as good old-fashioned storytelling can present fuller, more accurate, and I would argue more engaging views of the past. Conversely, quantitative data, when used in just the right, sparing amount, can complement traditional narrative history. And that quantitative data — wary as I am of math and my own math skills — has helped me to better understand the people called Cajuns and Creoles, and to convey that understanding to others.

Figure 5: 
Recapitulation of statistics
used in this essay
(click to enlarge)

ADDENDUM

After writing the above blog article about the limits of census data pertaining to Cajuns and Creoles, a couple of readers asked me to examine what the census results say about present-day language use in Louisiana. That is, how many persons in the state today speak French as their primary language in the home? This is indeed possible to gauge. But, as with census data concerning how many people identify as Cajuns and Creoles, we should examine the language data with a degree of skepticism. My review of these numbers, by the way, is hardly exhaustive, and is meant to afford only a glimpse at recent language traits in Louisiana, and of only two or three languages at that.

Let’s first go back a few years: in 2015 the U.S Census Bureau’s American Community Survey (ACS) 5-year estimate, accessible here, found 107,616 Louisianians speaking French “at home,” as well as 7,209 who spoke French Creole (now widely regarded not as a dialect of French, but as a distinct, standalone language called Creole, Kreyòl or Kouri-Vini). Unfortunately, later census results do not mention French Creole by any name, though perhaps stats for that language exist much deeper in the raw data. As for “French” in Louisiana, the U.S. Census Bureau regards that language as including more than one strain, including what it calls “Cajun” — despite the fact that some linguists now assert Cajun French does not exist (a view I myself do not embrace, but then, I am not a linguist).(14)

Figure 6: Table B16001,
“Language Spoken at Home. . . ,”
from the 2015 American Community Survey,
French and French Creole stats highlighted
(click to enlarge).

Seven years later the 2022 ACS 1-year estimate (the most recent year for which data is presently available, accessible here) found the number of French speakers in Louisiana to be only 64,302 — a decline of 40.24 percent since 2015. This trend might be explained by the dying out of older, primarily French-speaking Louisianians (and too few French-speaking youths taking their places, despite the advent of French Immersion schools). Even so, other explanations might account in part for this precipitous decline.(15)

Figure 7: Table B16001,
“Language Spoken at Home. . . ,”
from the 2022 American Community Survey,
French and French Creole stats highlighted
(click to enlarge).

Out of curiosity, I went to the 2020 ACS 5-year estimate, accessible here, which recorded the number of French speakers in Louisiana that year as 76,909 — considerably less than the 2015 figure of 107,616, but notably more than the 2022 figures of 64,302.(16) This 2020 stat, however, is consistent with a downward trend.

Figure 8: Table B16001,
“Language Spoken at Home. . . ,”
from the 2020 American Community Survey,
French and French Creole stats highlighted
(click to enlarge).

It is interesting to compare the number of present-day French speakers in Louisiana homes to that of present-day Spanish speakers. As elsewhere in the U.S., Louisiana has in recent decades experienced a significant rise in the number of Latino residents. (Fittingly, Louisiana belonged to the Spanish Empire from 1762 to 1800. It was thus a Spanish colony, at least on paper, when the Cajuns’ exiled Acadian ancestors began to arrive in the colony in 1764. Moreover, in 1779 Spaniards founded the south Louisiana town where I live, New Iberia, as the village of Nueva Iberia.)

According to the same 2022 table mentioned above, the number of Spanish speakers in Louisiana homes that year reached 157,029, while as mentioned the number of French speakers tallied 64,302 — the latter group being only 40.94 percent the size of the former. The ample number of Spanish speakers clearly exerts a real-world influence. From my experience, for example, it is rare to see French-language advertising in the state’s Cajun and Creole parishes. Ironically, it is more common to see Spanish-language advertising in traditionally French-speaking south Louisiana, whether for promoting churches, wireless service, or, as I noticed a few days ago while driving through rural Iberia Parish, a “venta de garaje” (garage sale).(17)

As with census stats concerning the per capita size of Louisiana’s Cajun and Creole populations, those concerning language use should be considered estimates — because that’s precisely what they are. Regardless, these figures can give us a feel for linguistic trends among the population and, when combined with historical, folkloric, and sociological methods, provide a more complete, more nuanced idea of Cajun and Creole ethnicity in the early 21st century.

 

Special thanks to Angela (Angie) Buchanan, Demographic Statistician, Ethnicity and Ancestry Branch, Population Division, U.S. Census Bureau; Alli Coritz, Ph.D., Statistician Demographer, Population Division/Racial Statistics Branch, U.S. Census Bureau; and Erik Hernandez, Survey Statistician, Social, Economic, and Housing Statistics Division, U.S. Census Bureau, for explaining the workings of more recent census data available on https://data.census.gov/.

Notes

*In the 1990 PUMS data I counted as “Cajun” both those who identified as primarily “Cajun” as well as those who identified as primarily “Acadian.” It could be argued I made a leap of faith by assuming that those persons of Acadian descent might be considered “Cajun.” Certainly, there would be exceptions. For example, a notable Acadian-derived population resides in Maine, which borders the former French colony of Acadie, now the Canadian provinces of Nova Scotia, New Brunswick, and Prince Edward Island. Those Mainers of Acadian descent most likely do not consider themselves “Cajuns” because Cajuns are largely a Louisiana phenomenon — the Cajuns arising in the first place from persons of Acadian descent intermarrying with other ethnic groups (French, German, Spanish, etc.) on the south Louisiana frontier. In general, however, I believed it reasonable to assume that persons in Louisiana (and east Texas, for that matter) who identified as Acadian by heritage might also identify as “Cajun.” Indeed, I would be surprised if they did not. Moreover, I regarded the percentage of exceptions to this rule as likely so small as to be irrelevant, because every other source of data (historical, journalistic, impressionistic, etc.) indicated the obvious: that a large population of persons residing in Louisiana and east Texas identified as “Cajuns.”

(1) Robert William Fogel and Stanley L. Engerman, Time on the Cross: The Economics of American Negro Slavery, revised ed. (New York: W. W. Norton, 1995). For an example of how quantitative data can be used to study Cajuns and Creoles from a sociological perspective, see, for instance, Carl L. Bankston III and Jacques Henry, “The Socioeconomic Position of the Louisiana Creoles: An Examination of Racial and Ethnic Stratification,” Social Thought & Research, 21 (April 1998): 253-277.

(2) Shane K. Bernard, The Cajuns: Americanization of a People (Jackson: University Press of Mississippi, 2003).

(3) “The U.S. Census Collection,” New York State Library, 8 February 2024, https://www.nysl.nysed.gov/uscensus.htm, accessed 22 May 2024.

(4) For more on Cajun GIs in World War II, see Bernard, “Cajuns during Wartime,” in The Cajuns, 3-22; and Jason P. Theriot, Frenchie: The Story of the French-Speaking Cajuns of World War II (Lafayette: UL Press, 2024 [forthcoming]).

(5) 1990 PUMS, for Persons. See Bernard, The Cajuns, xxiii-xxiv, 122-24, 152 (n. 9). See also Figure 1 in this blog article, from Shane K. Bernard, “Population Data, Cajuns,” Encyclopedia of Cajun Culture [author’s archived defunct website], 6 June 2002, https://web.archive.org/web/20020606133907/http://cajunculture.com/Other/populati.htm, accessed 16 May 2024.

For more on the interpretation of 1990 PUMs data in my book The Cajuns, see my blog article “Tracking the Decline of Cajun French,” 22 March 2011, http://bayoutechedispatches.blogspot.com/2011/03/tracking-decline-of-cajun-french.html#:~:text=For%20the%20first%20time%20%5B1946,of%20the%20century.%20.%20.%20, accessible here.

(6) Ibid.

(7) Bernard, The Cajuns, 120.

(8) “The Importance of the American Community Survey and the Decennial Census,” 13 March 2024, United State Census Bureau, https://www.census.gov/programs-surveys/acs/about/acs-and-census.html, accessed 16 May 2024.

(9) “People Reporting Ancestry,” United States Census Bureau, 2020, https://data.census.gov/table/ACSDT5Y2020.B04006?q=United%20States&t=Ancestry&g=040XX00US22,48, accessed 16 May 2024, or see Figure 2.

(10) Custom MDAT Table, for Creoles, based on ACS 5-Year Estimates Public Use Microdata Sample, 2020, https://data.census.gov/mdat/#/search?ds=ACSPUMS5Y2020&cv=ucgid&rv=ANC1P&nv=ANC2P&wt=PWGTP&g=0400000US22, accessed 16 May 2024, or see Figure 3.

(11) “People Reporting Ancestry”; Custom MDAT Table, for Cajuns, based on ACS 5-Year Estimates Public Use Microdata Sample, 2020, https://data.census.gov/mdat/#/search?ds=ACSPUMS5Y2020&cv=ucgid&rv=ANC1P&nv=ANC2P&wt=PWGTP&g=0400000US22, accessed 16 May 2024, or see Figure 4.

(12) Angela Buchanan, Demographic Statistician, Ethnicity and Ancestry Branch, Population Division, U.S. Census Bureau, [greater Washington, D.C., area], to Shane K. Bernard, New Iberia, La., 13 May 2024, email correspondence in author’s possession.

(13) Bernard, The Cajuns, xxiii-xxiv.

(14) Table B16001, “Language Spoken at Home by Ability to Speak English for the Population 5 Years and Over,” 2015: ACS 5-year Estimate Detailed Tables, https://data.census.gov/table/ACSDT5Y2015.B16001?t=Language%20Spoken%20at%20Home&g=010XX00US_040XX00US22_060XX00US2502706365&moe=true&tp=false, accessed 24 June 2024, or see Figure 6.

(15) Table B16001, “Language Spoken at Home by Ability to Speak English for the Population 5 Years and Over,” 2022: ACS 1-year Estimate Detailed Tables, https://data.census.gov/table/ACSDT1Y2022.B16001?q=B16001&g=040XX00US22_050XX00US36061$1400000, accessed 24 June 2024, or see Figure 7.

(16) Table B16001, “Language Spoken at Home by Ability to Speak English for the Population 5 Years and Over,” 2020: ACS 5-year Estimate Detailed Tables, https://data.census.gov/table/ACSDT5Y2020.B16001?q=B16001&g=040XX00US22_050XX00US36061$1400000, accessed 24 June 2024, or see Figure 8.

(17) Table B16001, “Language Spoken at Home by Ability to Speak English for the Population 5 Years and Over,” 2022.