Analysis and Reporting

A key consideration in many research studies is when to adjust for race, and how to interpret statistical models that do vs. those that don’t. For a summary of these issues, please refer to this citation. VanderWeele TJ, Robinson WR. On the causal interpretation of race in regressions adjusting for confounding and mediating variables. Epidemiology. 2014;25(4):473-484. doi:10.1097/EDE.0000000000000105
Brown Access here. 

1. Do not describe race using causal statements.Race and gender are often treated as individual attributes in inferential models. To treat these models as causal or inferential is a form of racial reasoning. In inferential models, unalterable characteristics cannot be a cause, but rather are statements of association between the values of an attribute and a response variable across individuals in the population (in some societies, race is alterable and considered a socially based characteristic) Only those variables that can conceptually be manipulated are eligible to represent causal processes. Thus, race is not a causal variable but rather an intrinsic property of the individual.

• Zuberi, T. (2000). Deracializing social statistics: Problems in the quantification of race. The Annals of the American Academy of Political and Social Science, 568(1), 172-185. Brown Access here. 

2. Avoid speculation beyond what the data support. For example, if an association is found between race/ethnicity and factors of interest, suggest researchers look at other research on this topic and put in the discussion how “behaviors do not exist in a vacuum” e.g., exposed to different stressors and may contribute to different patterns and behaviors. That is, provide some context for your results.  

• Ford, C. L., & Harawa, N. T. (2010). A new conceptualization of ethnicity for social epidemiologic and health equity research. Social science & medicine, 71(2), 251-258. Brown Access here. 

Cole, E. R., & Stewart, A. J. (2001). Invidious comparisons: Imagining a psychology of race and gender beyond differences. Political Psychology, 22(2), 293-308. Brown Access here. 

Research resulting in invidious (i.e., offensively discriminating) comparisons is characterized by five features: 

1. 1. 1. Dependence on an irrelevant norm (e.g., using one race/ethnicity as the implicit standard)

      2. Ordering differences hierarchically (e.g., attributes of one group are interpreted as evidence of some deficiency)

      3. Decontextualizing outcomes (e.g., disregarding context and other experiential factors)

      4. Comparing means without reference to distributions (e.g., making a distinction between group differences that are statistically significant but lack social meaning)

      5. Comparisons-based questionable operationalizations (e.g., using weak or dubious measures of the construct in question)

Link, B. G., & Phelan, J. C. (2001). Conceptualizing stigma. Annual review of Sociology, 27(1), 363-385. Brown Access here. 

Overarching goal of this review is to expand on the meaning of stigma and how it is conceptualized as the relation between an attribute and a stereotype. Four main components are discussed

1. 1. 6. Distinguishing and labeling differences: Avoid using language that relies on an oversimplification to create groups and be mindful that describing salient differences across groups differ dramatically according to time and place.

      7. Associating differences in negative attributes: Avoid linking labeled differences to negative stereotypes (e.g., someone hospitalized for a mental illness is viewed as being a risk for engaging in violent behaviors)

      8. Using language that separates “us” from “them”: Avoid using socially constructed labels to describe differences between groups of individuals (e.g., describing certain ethnic or minority groups as being fundamentally different kinds of people from “us”)

      9. Status loss: Avoid using language that links socially constructed labels to undesirable characteristics as this provides a rationale for devaluing, rejecting, and excluding groups of individuals

Bowleg, L. (2017). Intersectionality: An underutilized but essential theoretical framework for social psychology. In The Palgrave handbook of critical social psychology (pp. 507-529). Palgrave Macmillan, London. Brown Access here.. 

1. 1. Multiple identities/categories such as race and gender intersect at an individual level experience that can shape complex inequalities and at a larger social-structural level (i.e., racism).

   2. Acknowledging Intersecting identities/categories can be planned ahead of time in research designs with measures such at the Intersectional Discrimination Index (see Measures).

   3. Check out the Intersectionality Training Institute for intensive training opportunities. https://www.intersectionalitytraining.org/services/overview

   4. Example quantitative analytical approaches to intersectionality: Brown Access here.

   5. A review of current challenges: Brown Access here.

Jackson, J. W. (2017). Explaining intersectionality through description, counterfactual thinking, and mediation analysis. Social psychiatry and psychiatric epidemiology, 52(7), 785-793 Brown Access here. 

There is no quantitative approach that can include every dimension of intersectionality; however, some ways to investigate two intersecting identities/categories in a model can be with an interaction term or a step further with mediation analysis, for example.

Analysis of health inequities requires representation of multiple racial/ethnic groups. Recruiting diverse and representative samples is crucial. Involving the community is a key step in recruiting racial and ethnic minority (see Engaging Communities for more).

A review of studies that summaries  barriers and methods to facilitate recruitment of underrepresented groups is provided. Note that this is somewhat dated, and more recent summaries, particularly in specific population health areas more applicable to one’s interest, might be available and useful.] Yancey AK, Ortega AN, Kumanyika SK. Effective recruitment and retention of minority research participants. Annu Rev Public Health. 2006;27:1-28. doi: 10.1146/annurev.publhealth.27.021405.102113. PMID: 16533107. Brown Access here. 

Sample Size and Power

Cogua, J., Ho, K. Y., & Mason, W. A. (2019). The peril and promise of racial and ethnic subgroup analysis in health disparities research. Journal of Evidence-Based Social Work, 16(3), 311-321. Brown Access here. 

1. 1. This paper provides theoretical, methodological, and interpretation suggestions for conducting subgroup analyses. In addition to discussing issues pertaining to statistical power, this article provides recommendations for conducting exploratory and confirmatory racial and ethnic subgroup analyses. 

Bunn, V., Liu, R., Lin, J., & Lin, J. (2020). Flexible Bayesian subgroup analysis in early and confirmatory trials. Contemporary Clinical Trials, 98, 106149. https://doi.org/10.1016/j.cct.2020.106149 Brown Access here. 

1. 2. Bayesian methods of conducting subgroup analyses provide a potential framework to examine questions exploring differences across subgroups that are more robust and less susceptible to power concerns encountered with frequentist methods.