southern heritage and hate can’t be independent factors

L.J. Zigerell, a political scientist and frequent commenter here, has a new post on The Monkey Cage about attitudes towards the Confederate Flag among Southern Whites. Zigerell runs some regressions on attitudes data from 1994 and 2004 to show how attitudes towards the South in general and attitudes towards racial minorities predict support for the flag. Zigerell finds that even controlling for responses to racial attitudes, attitudes towards the South predict support for the flag. Here’s the conclusion:

Nevertheless, the results from these two surveys suggest a more qualified conclusion about the correlates of support for the Confederate battle flag. Surely racial attitudes reflect one such factor, but Southern heritage appears to be another.

This strikes me as wrong in an important way, but I’m curious if you all agree. Specifically, to me this conclusion rests on a “variables” conception of “heritage” and “hate”, where “heritage” means “responses to questions about love of the South” and “hate” means “responses to questions about attitudes towards racial minorities”. But that seems, well, wrong. It requires begging the question (in the sense of assuming the answer) that “Southern heritage” is a distinct thing that is separate from racism. But it isn’t. That’s not what racism is or how racism works. Zigerell adds some reasonable caveats about the age of the data, but I think misses this larger problem. The regression produces sensible seeming output, but the underlying constructs simply don’t make sense.

What do you all think?

Author: Dan Hirschman

I am a sociologist interested in the use of numbers in organizations, markets, and policy. For more info, see here.

22 thoughts on “southern heritage and hate can’t be independent factors”

  1. Hi Dan,

    Thanks for posting this. I’m interested in your thoughts and the thoughts of other commentators.

    You are correct that many and perhaps all “heritage” measures correlate with “hate” measures, but some of the “heritage” measures described in the post appear to be reasonably “hate” free, especially being able to name a Civil War battle, unless you want to argue that learning about the Civil War is inherently racist.

    Moreover, the “heritage” measure of having family who fought for the Confederacy seems to have a pretty clear causal order, unless a person’s racial attitudes can cause their ancestors to fight for the Confederacy; feel free to make a heritability argument to explain the family item, though.

    A common method for testing for separate underlying concepts in a set of data is factor analysis. I have posted the results of a factor analysis at my website suggesting that the racial attitudes largely load onto a different factor than the heritage items: Factor analysis isn’t perfect, but it seems preferable to simply claiming “But it isn’t” as a social science method.


    1. In the post, you list some of the questions you used for the Southern heritage variable:

      “Respondents were asked several questions that arguably captured feelings about Southern heritage, such as: whether the Confederate battle emblem reminded them of white supremacy and racial conflict or Southern heritage and pride; whether they were proud of what the Confederacy stood for; and how much what happens to Southerners will affect their lives.”

      My reaction was partly informed by those first two questions in particular. Asserting that the Confederate battle flag filled the respondent with Southern pride rather than reminding them of white supremacy strikes me as a perfect example of racism. Being proud of “what the Confederacy stood for” is racism – the Confederacy stood for slavery! I’m willing to believe that there is some statistical independence between those measures and explicit measures of something like support for interracial marriage. But that just means that racism is complicated and woven into complex ideologies of the past and present. Attitudes towards civil war history are racial attitudes, and are part of the structure of racism in the US.

      For an example from today, see these debates over the teaching of the Civil War in Texas.


      1. Hi Dan,

        Unless you are arguing that racism explains 100% of Southern heritage and support for the Confederate battle flag, I don’t think that you are disagreeing with the claim in the post that “[s]urely racial attitudes reflect one such factor.”


      2. I think Geoff’s point below captures much of what I was trying to get at. More generally, I think your approach implies that “racial attitudes” are some clean set of underlying beliefs well-captured by specific questions about attitudes towards discrimination, etc., and similarly, “Southern heritage” is some real set of underlying beliefs well-captured by specific questions. But I don’t think that assumption is reasonable: Southern heritage, as a discursive construct, is enmeshed in a discussion about race. It is an essentially racial construct.

        I think your findings are useful and interesting, and in some sense unsurprising. People who were willing in 2004 to say that interracial marriage should be banned/discouraged are a different sort than people who reject that attitude but still think that the South was built on something other than slavery, and that contemporary racial inequality has little to do with slavery or segregation since those problems are in the past, and safely separable from the good part of Southern heritage.

        Put differently, looking at explicitly racist answers to explicit questions about race picks up one kind of racism (a kind on the decline but still sadly existent), looking at racist answers to questions about Southern history picks up a different kind of racism, one that has managed to masquerade as not being about race while justifying contemporary racial inequality.


      3. Hi Dan,

        “…looking at racist answers to questions about Southern history picks up a different kind of racism…”

        Being able to name a Civil War battle is a racist answer?

        For what it’s worth, I just constructed a scale based on the Southern Focus Poll items about word choice for food: coke (vs. soda and other terms), skillet (vs. frying pan and other/don’t know), dressing (vs. stuffing and other/don’t know), and supper for the evening meal (vs. dinner and other/don’t know). The Southern word choice scale associated with support for official use of the flag (but not with opposition to prohibition of the flag for private use) at conventional levels of statistical significance, controlling for the lone measure of racial attitudes in the survey (support for segregation).

        I then did a factor analysis with the three heritage items and one hate item discussed in the Monkey Cage post, plus the four food word choice items. The three heritage items fell on the first factor, the supper and segregation item fell on the second factor, and the three remaining food word choice items fell on the third factor. I’m not sure what kind of racism that third factor is tapping.

        The measures of “heritage” and “hate” in the data correlate with each other, cross-sectional data do not permit causal inference, and the statistical approach that I used was not perfect. But it was the same approach that Piston and Strother used to argue that heritage had no aggregate independent effect. And I added factor analyses in response to claims that the heritage measures were proxying racial attitudes. I’m not sure what else can be done with the data that would be as convincing.

        Maybe the patterns that I found in the two publicly-available datasets that I uncovered do not hold in contemporary times. Maybe these data were outliers in 1994 and 2004. Maybe I’m not analyzing the data correctly. But at least I’m analyzing data. There is no way of knowing whether your claims are correct unless the claims are subjected to empirical analysis.


      4. I don’t know that this will help our dialogue much, but I have very similar problems with Piston and Strother’s framing. I’m not arguing that you’re wrong and they are right, I’m arguing that survey data like you (both) are analyzing can’t pick up and disentangle the relevant constructs here because they are inextricably linked. We can come up with a new construct called, I don’t know, “love of Southern food culture and linguistic quirks” and show that it is (or isn’t) linked to support for waving the Confederate battle flag. But for many involved in the debate, that flag is Southern heritage. That’s exactly what’s at stake (or at least part of it).

        “Being able to name a Civil War battle is a racist answer?”

        As I said before, my reaction was premised on the initial post which listed among the questions analyzed “whether the Confederate battle emblem reminded them of white supremacy and racial conflict or Southern heritage and pride” and “whether they were proud of what the Confederacy stood for.” To me, these questions are transparently not measures of some de-racialized “Southern heritage” (if such a thing is even possible), but very clearly questions about how the respondent interprets the history of Southern oppression of its black members. Just because the questions lump together statistically (and separately from explicit measures of racial attitudes) doesn’t tell us anything about whether they map substantively onto what we care about.

        At to “at least I’m analyzing data…”, I think trying to answer the wrong question leads to bad analysis no matter how much care you take with the data and analysis. Your question – and Piston and Strother’s – begins from a false premise (that “racial hate” and “Southern heritage” are reasonably treated as separable, individual-level attitudinal constructs), and no amount of survey data can fix that.

        Liked by 1 person

      5. Hi Dan,

        I think that I understand your argument better.

        Your argument appears to be rooted in the claim that “racial hate” and “Southern heritage” are not separable (“inextricably linked”). But I don’t know of any evidence indicating whether this claim is correct: the claim appears to be only an assertion.

        I’d agree with a more modest and generalized version of your argument, that it cannot be known for certain whether statistical control in observational research isolates the effect of the explanatory variable of interest. But that argument also appears to apply to research on such questions as the effect of CO2 on global temperatures and the effect of race in criminal justice outcomes.

        In that case, I’d argue that observational research with statistical control can have positive value based on the patterns in the data, the quality of the data, the quality and extent of the statistical control, and other elements of the research design.

        For example, if a heritage/flag model has high-quality racial attitudes measures that reflect all the different manifestations of racial attitudes, I think that it makes sense to update beliefs about the effect of heritage differently based on whether the residual effect of the heritage measures is very large or zero.

        (The ideal mechanism by which to test whether statistical control in observational research isolates the effect of the explanatory variable of interest is the ability of a model to make accurate predictions for a wide range of inputs, but attitudinal survey research doesn’t lend itself to the prediction-testing business very well.)


  2. It’s telling you ask what you all think rather than y’all! Does it to some degree depend upon how we understand the relative adaptability of cultural structures? BBQ, Southern Gothic Literature, blues, etc. are not by necessity racist, unless somehow being part of Southern Heritage infects them. Something like “Southern Heritage” with all its historical components might well be, on some level, necessarily racist, in the way that something like, say the Church of Latter Day Saints might be necessarily homophobic. Mormons are a complicated example, because there’s an actual institution at play, but there are lots of gay or feminist Mormons that accept many of the cultural structures without accepting all of it, or even the power of the institution. So, on one hand, I’m entirely sensitive to the critique that many Southerners simply do not know their own history. But I’m also a bit wary of the idea that a certain Southerner–or member of any community (say, the imperialist and neocolonial United States?)–can’t actively reinterpret the rather large cultural structure that is “Southern Heritage”. Now, of course, just because person (A) interprets the South as not racist doesn’t mean that that interpretation can undo the history of Southern Heritage, nor does it mean that person (B) is wrong to be offended by the entire idea of Southern Heritage (in the same way, say, that someone might interpret any effort at being Mormon to be necessarily a homophobic enterprise). Yet I’m uncomfortable telling a thoughtful southerner proud of the culinary, musical, and literary tradition of the region that any such pride is by its nature racist.

    Liked by 1 person

  3. I also think Dan, that in your response, you’re conflating two things: the idea of Southern Heritage writ large and the specific claim the Confederate Flag is about Southern Heritage (not hate). While the former is a much bigger and complicated mess (though still obviously sensitive to the critique of inherent racism), the latter is what the article itself was really about, and that’s where I think I agree with you a bit more, although I will say that symbols are pretty malleable and can pretty easily take on multiple meanings to multiple communities. The rewriting of the history of the Confederacy was incredibly effective (having lived in the South I can vouch for this): I think there’s a lot of Southern whites who are wrong about the heritage-not-hate interpretation of the flag but who do honestly believe it.


  4. To me, buying into the heritage-hate dichotomy means you’ve more or less already accepted the argument for the tacit social acceptance of racism of the kinder and gentler “racism without racists” variety.

    There’s a certain caricature of racism that fixates on lurid episodes of white sadism towards blacks when the more widespread and pernicious form of racism has always followed logically from the clear “upside” to racism: the indisputable privilege that whites have of benefitting in virtually every possible way at the expense of blacks without even having to shoulder the emotional burden of thinking about it that way.

    Certainly, in moments of crisis for white supremacy (eg. reconstruction, the civil rights movement) the more sadistic variety is bound to flare up, but I think it’s historically naive to assume that white people have ever spent that much time thinking about the consequences of their actions for other races and their outsized ability to inflict violence and terror through their actions–symbolic or otherwise.


  5. This reminds me of the racist (white, rural, Maryland) white man I overhead at the fireworks show. These are my contemporaneous notes:

    Two white guys at fireworks show (crowd 1/2 white, 1/4 black, 1/4 hispanic). Guy talking hasn’t been able to get off work to see fireworks for the last six years. Getting dark. Other guy asks what he thinks of the whole Confederate Flag thing. “It’s ridiculous. They don’t understand it. They don’t understand what the Confederate Flag meant. Half the people don’t understand half the stuff about this country. Hell I fly one in my front lawn. Just like I put a big Merry Christmas thing out at Christmas. If you don’t like it come talk to me. … Half my family fought for this county. [… Inaudible …] Well, on my step-mother’s side one guy fought for the South. But most of my family is from northern states.”

    He’s got the racism connection, just missed the Southern thing.

    (Note on my own racism: when I developed the pictures I took and could see the crowd better I could tell it was quite a bit more than half white.)


  6. Dan, you’re right that Southern Heritage (all of it) is an essentially racial construct. Slavery has been a part of Southern heritage from its beginning. But that’s also true of the United States, and how we interpret that discursive construct, as you put it. So what’s interesting to me is the question of that “heritage” and how folks interpret it. Now, again, there are plenty of people who would say they care about “Southern heritage” and completely hate the flag. So that’s important to separate. But, even regarding the flag: how does any flag–Confederate or American–work as a text? Who controls the interpretation? Who is allowed an interpretation? What to do when one meaning (and the historical and most dominant one) causes real harm? And how much does the original intention of that text (which was, to be clear, opposition to the civil rights movement and white supremacy) matter if people now view the text as wholly different? And I think Geoffery’s obviously right that much of this has to do with white privilege. Whites have the power to interpret the flag in any way they choose because the difference between a racist or a not racist interpretation has no real stakes for them. Yet the point is, whether or not the interpretation is racist, it is necessarily racialized (something they often don’t see: but, to be fair, neither do white folks in many regions of the country). African Americans have no such autonomy of interpretation: the racialized nature of the flag makes it immediately racist as well. But as various folks here have made clear, that difference in autonomy of interpretation has more to do with larger structural and historical questions than individual attitudes on race. Which is to say, the very fact that such a question could be posed shows that heritage vs. hate is a false dichotomy: the ability to say it’s just about heritage is premised upon a structural inequality rooted in hate.


  7. Another approach to this question is to do a series of counter-factuals. Could “southern heritage” survive without racism? We know for example that the flying of the confederate battle flag in southern states was a response to the Civil Rights movement. Another interesting one, is since almost all African Americans have contemporary or recent (at least compared to the Irish and Irish heritage), how do they score on the southern heritage scale?

    Finally, a different approach is to take this not as a variable/correlation question, but as a configurational question. Create bins for your key variables, add other interesting conditions like race or region and do a cluster analysis. What proportion of your samples are in each bin? What configurations don’t exist? Does heritage exist in the absence of racism? Correlation based evidence tends to mimic shifts in the middle of the distribution, but ignores both configurations and extremes.

    Liked by 1 person

    1. Hi Don,

      Good ideas. I estimated a few models based on your suggestions. I don’t have time right now to double check the models and draft a summary, but I plan to post a comment later tonight or sometime tomorrow.


    2. Hi Don,

      I am not familiar with cluster analysis, but here are results for samples limited to black respondents and for samples limited to white respondents who approved of black-white marriage (Hutchings and Walton survey) or who supported integration (Southern Focus Poll). I used the same model setups for these analyses as were used for the reported analyses in the blog post.

      Hutchings and Walton survey, black respondents only (n=165).
      The “reminds of” heritage item associated with black preference for the flag at p<0.05 (two-tailed test) when each racial attitude item was placed in a model one-at-a-time and at p=0.058 (two-tailed test) when all three racial attitudes items were placed in the model at once. The "pride" heritage item associated with black preference for the flag at p<0.05 (two-tailed test) when each racial attitude item was placed in a model one-at-a-time and when all three racial attitudes items were placed in the model at once. Black respondents were not asked the "close to Southerners" heritage item.

      Southern Focus Poll, black respondents only (n=123).
      Heritage measures were entered one-at-a-time into a model with the segregation item. There was no statistically significant effect for black respondents for the importance of the Civil War or for having family who fought for the Confederacy. Being able to name a Civil War battle associated with *less* support for official use of the flag at p<0.05 (two-tailed test). Having family who fought only for the Confederacy (and not the North) also associated with *less* support for official use of the flag at p<0.05 (two-tailed test), but only one black respondent in the poll reported having family who fought only for the Confederacy. Statistically significant patterns for black respondents were reversed for opposition to prohibition of private use of the flag: more Civil War battle knowledge and having family who fought only for the Confederacy correlated with more support for private use of the flag.

      Hutchings and Walton survey, white respondents who approved of black-white marriage only (n=192).
      The “reminds of” and “pride” heritage items associated with preference for the flag at p<0.05 (two-tailed test) when each racial attitude item was placed in a model one-at-a-time and when all three racial attitudes items were placed in the model at once. The "close to Southerners" heritage item did not associate with preference for the flag at p<0.05 (two-tailed test) when each racial attitude item was placed in a model one-at-a-time or when all three racial attitudes items were placed in the model at once.

      Southern Focus Poll, white respondents who supported integration only (n=339).
      Heritage measures were entered one-at-a-time into a model alone. Only the importance of the Civil War item reliably associated with support for official use of the flag at p<0.05 (two-tailed test); the item for family who fought only for the Confederacy had a p-value of 0.101. The importance of the Civil War item, the family-in-the-Confederacy item, and the family-only-in-the-Confederacy item associated with support for private use of the flag at p<0.05 (two-tailed test).


  8. Late to the party, but this debate reminds me of the Redskins debate as it seems to me to have exactly the same character. Those who support the Redskins logo insist they just love football and have hometown pride and deny the meaning of the name. Given the whitewashing of US history in most people’s education and the extent to which students are taught that the Civil War was not about slavery and are taught an expurgated version of US history, it is not surprising that they might well associate the Confederate flag with things other than slavery and White supremacy. But that does not mean that the flag is not a symbol of slavery and White supremacy, any more than White football fans’ ignorance means that the Redskins name and logo are not racist symbols. The Nazi flag is a symbol of Naziism whether or not a person knows what the symbol means, and the Nazi flag is just as much as symbol of Naziism even though there were lots of anti-Nazi Germans and even though people can have pride in Germany and have fought for their country without believing in genocide. In Germany, everybody understands that the meaning of the Nazi flag is Naziism, not generic German patriotism. This has to do specifically with how history is taught in Germany which, in turn, has to do with losing the war and who wrote the history.

    The US Civil War is the only case in which the losers wrote the history.

    How about a study of the correlation of support for the Confederate flag with a question about whether the South seceded to save slavery? My prediction is that you’d get a strange pattern of results, with overt White supremacists knowing the history and supporting the flag, self-conscious anti-racists agreeing with the history question and opposing the flag, and the great uneducated middle denying the slavery connection and having support for the flag that is tied to regionalism or whatever.

    I was just having a debate with a relative over this: his great-great-grandfather was an anti-slavery Southerner who fought on the side of the South in the war, and he was trying to argue that this connection meant that the flag didn’t represent slavery. Trying to get people to see that individuals might fight in a war on a side that don’t necessarily agree with out of patriotism or regional loyalty is tough, and trying to get them to distinguish personal motives from the meaning of a national symbol is even harder.

    And to cycle back, I do think that the sociological processes are not that dissimilar in the Redskins and Confederate flag symbols.

    Um, should say that this is not to disparage the value of an individual-level analysis of support but important to distinguish that from debates about what a symbol “really means.”

    PS I have not fact checked, but I’m pretty sure the Confederate Flag was not added to Southern state flags until the Civil Rights era, which (if true) makes it even more clear that its meaning now as well as in the past is as a symbol of White supremacy.

    Liked by 1 person

    1. In Mississippi at least, the adoption of a Confederate-like flag dates to the 19th century. My understanding is that the drive to fly the original battle flag dates to the Civil Rights era (NPR puts it this way: “The flag was first flown over the [South Carolina] state Capitol dome (passed by the Democratic Legislature) in 1962 to mark the centennial of the start of the Civil War, but many saw it as a reaction to the civil-rights movement and school desegregation.”).


    2. Hi OW,

      Your post seems to assume that a symbol has only one meaning and that this meaning can be divined. It seems more reasonable that a symbol can have multiple meanings imputed to it and that these meanings can change — and can be changed — over time. That’s what the early Christians did to a lot of pagan symbols, after all.

      It seems that conducting a survey to determine what the Confederate flag “really means” would be as useful as conducting a survey to determine whether an exclamation point really is punctuation or factorial indicator. The fact that most people might be unaware of the exclamation point’s use in mathematics does not mean that the exclamation point is not a mathematical symbol.


  9. Northern democrats of the 20th century, in large northern cities via vice and broken windows policing, housing policies and subsidies, teacher’s unions, have done as much to harm blacks as southern democrats did to harm blacks in the 18th and 19th centuries via slavery. Northern cities continue to have the worst problems with segregated schools and neighborhoods in the country. Saying racism is irreducible from affections for the South is an analytical cop out, just like the rest of the “keep your regressions off my critical humanism” complaints.


Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.