This Academic Paper on Biracial Children with African-American Fathers is Likely a Fabrication
Update: 11/17/2018: It appears that SSRN has removed the Calloway paper from their database.
In July of 2015, self-described “independent” researcher Tiffany Calloway published the paper “Ninety Two Percent: Examining the Birth Trends, Family Structure, Economic Standing, Paternal Relationships, and Emotional Stability of Biracial Children with African American Fathers” at SSRN (The Social Science Research Network). The paper was apparently never peer-reviewed, nor was it published in any journal that I am aware of. The paper has not been discussed in the social sciences literature, but it has received some attention on internet forums, especially those forums associated with hate groups and/or anti-black bias. A reading of the paper, however, indicates many problems, including inappropriate statistical methods, highly questionable results, and outright plagiarism. In my opinion, it is extremely likely that the survey detailed in the paper was not conducted as described or was never conducted at all.
Methodology of the Survey
The author claims that a survey was undertaken among 1,000 women of “Hispanic, Asian and Caucasian” origin who had given birth to children with African-American fathers. The survey group was divided equally between the ethnicities. Oddly, the sample size and the racial/ethnic distribution were not stated in the body of the paper and appear only in the abstract. In the “Research Methods” section of the paper, the author notes that the participants in the survey hailed from across the United States and that they were recruited via “media outlets such as newspapers, internet, and radio” and that “ most of the respondents were recruited through internet advertisements.” It is obvious that the author, who lists herself as independent, incurred at the very least modest expenses in placing these advertisements across the country. The paper further describes a process of documenting and verifying respondents that would be time intensive — but by no means impossible — for a single researcher.
The percentages and their discussion appear to be the author’s original work. However, there are several instances of plagiarism throughout the paper. In the introduction, several sentences are copied nearly verbatim from sources not cited by the author. For example, the partial sentence “boys raised in a single-parent household were more than twice as likely to be incarcerated, compared with boys raised in an intact, married home” appears in a 2012 article in Slate.
More troubling, in the section marked “Emotional Effects of fatherlessness,” all five numbered sentences are plagiarized from a single web page titled “Statistics of a Fatherless America”. This web page is not cited and the paper’s entire bibliography consists of sources corresponding to the five sentences in the “Emotional Effects” section.
The paper is filled with anomalous, sometimes even incredible, results. These results might seem possible with a lower number of respondents, but the survey group of 1,000 renders the results extremely unlikely. In her discussion section, the author claims that “at the time of the study 100% of the women were single and unmarried.” With a sample of 1,000 mothers of interracial children, finding not even one who was married strains credulity. Even if somehow the selection method was biased in favor of divorced women, the result of zero married women would be highly implausible.
One could (unconvincingly) argue that perhaps this particular demographic might be extremely unlikely to be married at all. But the survey finds that 17% (170) of the women had been married to the father of their child at some point. By the time they were surveyed, all 170 women reported being divorced from their child’s father and that they had not remarried.
There are also several implausible results in other tables. For example, the percentage of Caucasian women who report that their child’s father is active in the child’s life is 3%. Yet, 6% of Caucasian respondents report that their child has a good relationship with his/her father. Finally, while not a statistical oddity, some tables divide the data in bizarre ways. Table 3, “Number of children that the participant has,” breaks the data into columns for “1,” “2–4,” “5–8,” and “9+” children. This is true even though no respondent reported nine or more children. Table 5, “What are your annual earnings?” includes one column marked “Unemployed” and another marked “0–10,000.” It is unclear how respondents would be divided here.
Flaws in Methodology
The author describes how the data were handled: “The data amassed was broken down into percentages, and the individual percentages were averaged.” It is unclear why the author chose to use this non-standard method. Since the sample groups are purportedly equal, averaging percentages will give accurate results. However, the author also asks questions of subsets. For example, the author asks a follow up question of the 10% of women who responded “Yes” to the question “Is the father active in the child’s life?” But since the sub-sample has an uneven racial/ethnic makeup, averaging the resulting percentages will introduce error. The actual errors introduced into this paper appear to be small — on the order of one or two percentage points.
At best, this paper is an extremely sloppy piece of research. However, the blatant plagiarism indicates a tendency for the author to engage in unethical research practices. This author appears to have no other publications, and it seems extremely unlikely that an independent researcher would spend so much time and expense on a survey only to plagiarize non-essential elements of the paper, thereby calling the accuracy of this time-consuming effort into question. Further, the author appears to employ nonsensical statistical procedures (e.g. the averaging of percentages). Finally, although by no means conclusive, the paper contains numerous grammatical errors and awkward constructions that are seldom the hallmarks of a quality researcher. It should be noted that there is no contact information available for Tiffany Calloway at the SSRN website, so it was impossible to contact the author about this paper.
Given the size of the sample, the lack of details about the survey, and the repeated plagiarism, the most likely conclusion is that the survey described in this paper was fabricated. There is little post-publication discussion of this paper online, but the vast majority of mentions are associated with white nationalist groups or other hate organizations. The paper was posted on July 4th, 2015 and one of the first places that mentions it is a post on the neo-Nazi site Stormfront on July 10th. Interestingly, this post links to a now-defunct GoFundMe page under the author’s name, suggesting a possible monetary motive:
Unfortunately, the questionable results of this paper are still repeated in non-academic discussions. The article is featured at a site called “Library of Hate,” and it has resurfaced in each of the last three years in many different internet forums. Since 2015, the paper has been downloaded nearly 7,000 times and the abstract has been viewed more than 71,000 times. These results give it a rank of 3,425 in the SSRN database of over 800,000 articles — well within the top 1%.
Interestingly, SSRN appears to have removed this article from its own search results. This may be because in 2017 I alerted SSRN to some of the problems mentioned here, including the plagiarism from the introduction. However, there is no notice on the paper’s page at SSRN and the paper still appears in a Google Scholar search.
Given the plagiarism discovered in this paper, the questionable study design and statistical anomalies, and the paper’s direct and rapid spread to internet forums specializing in hate speech, it is extremely likely that this paper was fabricated in order to be used as propaganda against African-Americans in general and interracial relationships in particular. Slipshod research methodology, unethical research methods and a hoax perpetrated for monetary gain — or some combination thereof — cannot be ruled out entirely.