Abstract
Objective: Prolonged experiences of discrimination can be a major source of physical and health-related stress, particularly in older Black Americans. However, there is limited information on the relation between discrimination and anxiety, particularly within the context of other constructs that influence the manifestation of anxiety symptoms. For example, several studies have suggested that ethnic identification may provide psychological and social resources to deal with the effects of discrimination. This study aims to further understand these processes.
Method: This study utilized structural equation modeling (SEM) to examine predictors of anxiety symptom severity in a sample of African American and Afro-Caribbean adults aged 55 and older from the National Survey of American Life (N = 1,032).
Results: The final structural regression model revealed acceptable fit indices, and was successful in measuring latent anxiety symptom severity, showing that more experienced discrimination was related to higher anxiety and anxiety was directly related to mental health rating. While higher experience of discrimination was associated with higher anxiety, it was not directly related to mental health rating. However, contrary to expectation, ethnic identification did not serve as a protective factor between experienced discrimination and anxiety. As individuals aged, they experienced less discrimination and reported poorer self-rated mental health.
Conclusions: While age served as a protective factor for perceived discrimination and anxiety symptom severity, ethnic identification did not. Implications for those working to reduce anxiety symptoms among Black Americans are discussed.
Disclosure statement
The views expressed herein do not necessarily reflect the views of the Durham VAMC.
Data availability statement
The authors of this manuscript have not published this data previously and do not have any working papers, conference presentations, theses, or dissertations using this data. The data reported in this manuscript were obtained from publicly available data known as the National Survey of American Life (NSAL), which is a subset of the Collaborative Psychiatric Epidemiology Surveys (CPES) and can be found here: https://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/20240/summary. A bibliography of journal articles, working papers, conference presentations, and dissertations using the NSAL is available at https://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/20240/publications. The variables and relationships examined in the present article have not been examined in any previous or current articles, or to the best of our knowledge in any papers that will be under review soon.
Notes
1 Model fit statistics were computed based on the scaled chi-square. The model was estimated using Procedure CALIS in SAS/STAT 14.1.
2 The 95% CI was obtained using PRODCLIN (MacKinnon et al., Citation2007).
3 Full Information Maximum Likelihood (FIML) is another option to deal with data missing at random. However, FIML assumes multivariate normality, which appeared untenable with this sample. We therefore opted for Multiple Imputation.
4 The fully conditional specification (FCS) method (Brand, Citation1999; Van Buuren Citation2007), as implemented in PROC MI, was used for data imputation. This method assumes a multivariate joint distribution, but not necessarily multivariate normality. PROC MIANALYZE was used to obtain final estimates and standard errors across 15 imputed data.