I get really annoyed when someone posts about some research, like this great post on Alas, a Blog on race and low birth weight, among other things, and someone who has no clue about research says “oh, but what about (insert obvious confounder that was obviously controlled for by the researchers here). Therefore, this study I didn’t even read must be total BS!”.
Some racism apologist commenter (RonF) decided to say that a study on the influence on racism on low birth weight is based on perceived racism, not real racism, and besides, he’d really like to see some research on poverty and low birth weight because that’s totally the real and only association with low birth weight.
Here is my reply:
Thanks for a great post.
Ron, you know, researchers who get published in major publications actually have to analyze their data for obvious confounders. (That’s a fancy statistics term for other variables that may be responsible for the outcome, like poverty instead of race being associated with low birth weight). I am sorry if I sound sarcastic, but this is a pet peeve of mine.
In fact, these same researchers wrote another entire study about poverty, race and low birth weight. You said you’d really like to see research on that topic. Do you actually believe you are the first person to consider this connection, and that you wouldn’t be able to find it? Or do you mean you’d like someone else to look it up for you, and in the meantime you’d just like to muse about the harms of racism being faked in published research by whiny black mothers who are mistakenly perceiving nonexistent racism and researchers with guilt and poor analysis skills?
It’s really frustrating when someone hasn’t bothered to read any of the abundant research that shows that race is an independent risk factor (independent from income and social economic status) for all sorts of health care outcomes, including low birth weight, but feels qualified to say the outcomes are incorrect and they have a much better theory, based on seeing no data and no research.
A simple search on the authors if you were trying to read the study before criticizing it, or on the topic of race, poverty and low birth weight before hypothesizing about it would find this:
Here is the abstract:
Objective To determine whether women’s lifelong residential environment is associated with infant low birth weight. Methods We performed race-specific stratified and multivariate binomial regression analyses on an Illinois vital record dataset of non-Latino White and African-American infants (1989–1991) and their mothers (1956–1975) with appended United States census income information. Results Non-Latino White women (N = 267) with a lifelong residence in low-income neighborhoods had a low birth weight (< 2,500 g) incidence of 10.1% vs. 5.1% for White women (N = 10,647) with a lifelong residence in high-income neighborhoods; RR = 2.0 (1.4–2.9). African-American women (N = 18,297) with a lifelong residence in low-income neighborhoods had a low birth weight incidence of 17% vs. 11.7% for African-American women (N = 546) with a lifelong residence in high-income areas; RR = 1.5 (1.2–1.8). The adjusted population attributable risk (PAR) percent of LBW for lifelong residence in low-income neighborhoods was 1.6% for non-Latino White and 23.6% for African-American women. Conclusions Non-Latino White and African-American women’s lifelong residence in low-income neighborhoods is a risk factor for LBW; however, African-Americans experience a greater public health burden from this phenomenon.
Translation: African-American women who have lived in high-income neighborhoods had worse birth weight outcomes than white women who lived in low income neighborhoods.
There has been plenty of research that simply being a minority in this country is enough to affect you in many significant ways. It doesn’t matter if someone on a website hypothetically believes minorities have ever experienced “real” racism to make the measurable effects of racism true.
Edited with an update. The fool continues to defend institutional racism by grasping at straws in his effort to criticize the study, which he STILL HASN’T READ. Laughingly, he thinks there is a problem with the N (the number of subjects in the study.) I was forwarded research by a classmate yesterday of a gyn medical device that one of our professors is a fan of yesterday. The largest study had an N smaller than 30. The study I link to above? TENS OF THOUSANDS. Apparently this dipshit thinks researchers are supposed to misrepresent what happens in real life (like, more African American women live in persistent poverty in this sample, and lots more of them have low birth weight (LBW) infants) to make the numbers match exactly between groups (and therefore…prove nothing?) Then, Mr. Concern Troll says it is all well and good to talk about the lofty goals of eliminating racism, which he is not denying (except that he is) but realistically, what are we supposed to do about this?
Here is my reply:
RonF, you need to read more than an abstract to know what was controlled for. Also, in a multivariate analysis in which the researchers look for many risk factors, as this was, researchers may not even choose to publish risk factors that did not have clinical significance.
And, a disparity in the N numbers is not a problem with research, especially if one of the groups is a minority and is naturally present in lower numbers. In fact, the N numbers are NOT that disparate in this study, and just guessing that is true does not make it true. In fact, they clearly prove a higher prevalence of LBW in African Americans.
It’s amusing in a sick way, because the N is one of the strongest parts of this study. I am having a hard time having this discussion with you and not totally calling you out as a rabble rouser grasping at straws to apologize for and diminish racism.
If you want to discuss the fine points of statistical analysis, um, read a whole study first, and then take a biostats class.
What is important is the power of your N number. And, the power of this study is impressive. If you knew anything about research, or even read the full text of any of these studies, you would know that.
What else must be done? Well, first of all, we have to get ignorant white men to stop denying facts about the extent of the problem on websites so we can have a productive conversation about this.
Edited again to add:
Someone came on to defend RonF, albeit she claims it is a reluctant defense. She also didn’t bother to read the original research or the other study by the authors I linked to, but thinks she is qualified to comment on their flaws. She criticizes the N and the “statistical analysis” of the original qualitative study. Here is my answer:
The statistical rigor I was referring to was of the quantitative research done by the same authors, which I link to in my very first post and from which I pasted the abstract. There is only one “N” in the qualitative interview, since there is only one group of subjects, so I assume RonF was also referring to the quantitative study when he claimed that there was a “disparity” between numbers in multiple groups.
You don’t control for confounders or variables in a qualitative study with interviews. It is not appropriate, for obvious reasons, other than in your subject selection. The qualitative research was done with a typical number of subjects for qualitative research, a small group, and is not set to the same “rigor” standards as quantitative research.
In other words, there is absolutely no statistic analysis in a qualitative study, so criticizing a qualitative study for its statistical analysis when there isn’t any, is, well, a sign you have no idea what you’re talking about. In fact, a qualitative study that tries to assign quantitative values to open ended interview answers is seriously flawed and should be criticized for even attempting statistical analysis, since the study method is not suited for statistical analysis.
Qualitative research is usually open ended interviews with a small group of subjects to get more nuanced information about complicated, multi factorial topics. Like racism, which is obviously sadly lacking in nuance in much of the discussions of the topic. It is a common technique in health issues that also involve power balance questions, such as pregnancy and birth.
As for “racial discrimination”, I am really missing the finer point here. If you show me flaws in so called statistical analysis of all of their background literature review, including the excellent quantitative study with the huge N, that point to simply being African American as being a risk factor, one that is greater than genetics or poverty or whatever other risk factors are examined, then we can talk.
It seems to me, yet again, as you are linking to the layperson’s news article that discusses the scientific article, that you, like RonF, have not bothered to read any of the original research. I would really think twice about discussing “rigor” when that is your method of looking into a study’s quality.
I know I am coming across as really pissy, and I apologize, but I would never go on a website and pretend to criticize something as technical as statistical analysis of medical research if I didn’t have a pretty good idea that I had an accurate criticism. It would be like me going on a website on engineering and start telling people their blueprints are messed up because I read some other person’s paragraph about their blueprints. It’s more complicated than all of that, and this armchair amateur hypothetical musing is one of my pet peeves.