ABSTRACT
How do American men’s attitudes about masculinity differ across intersections of race/ethnicity, immigration status, and education? This paper uses the NSFG 2011-2019, a large survey (n = 17,944) representative of American men aged 15-44. It analyzes white men; Black men; non-immigrant Latinos; and immigrant Latinos, with each broken down by less than a bachelor’s; a bachelor’s degree; or an advanced degree, for a total of 12 intersections. Most differences between men of different races/ethnicities/immigration statuses were between men with less than a bachelor’s. Several groups were more conservative on some attitudes but not others. For instance, among men with less than a bachelor’s, white men were more conservative than Black men regarding an attitude about going to the doctor, but less conservative than Black men on attitudes about showing pain or men’s sexual needs. Additionally, the attitudinal differences that emerged were distinct for different levels of education. Among men with less than a bachelor’s, most significant differences emerged regarding the attitudes about going to the doctor and men’s sexual needs. In contrast, among men with a bachelor’s, most differences emerged regarding the attitude about showing pain. Among men with the same racial/ethnic identity and immigration status, men with lower levels of education were more likely to endorse conservative attitudes about masculinity. All three intersections are meaningfully related to attitudes about masculinity, and future research about masculinity attitudes should not analyze social identities/statuses separately but rather as they intersect with one another.
Acknowledgments
Thank you to the Sexualities Project at Northwestern (SPAN) for providing the postdoctoral fellowship that allowed me to begin work on this project.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Ethics approval
Ethics approval was not necessary for this paper because it used publicly available de-identified data.
Informed consent
The National Survey of Family Growth obtained informed consent from all participants.
Financial interests
The author has no relevant financial or non-financial interests to disclose.
Availability of data
All National Survey of Family Growth data is publicly available.
Code availability
Stata code is available upon request.
Authors’ contributions
Tony Silva was responsible for all data analysis and manuscript preparation.
Notes
1. In the United States, most studies combine measures of race/ethnicity into a single variable with mutually exclusive options: for instance, non-Hispanic white, non-Hispanic Black, Hispanic, and non-Hispanic other (these four classifications were the only ones available in the NSFG, the survey this paper uses for data). Race and ethnicity are understood very differently in other countries and regions of the world.
2. Different countries and regions have different terminology for education. By ‘primary and secondary education,’ this paper refers to education under the level of college or university (i.e. from childhood to ages 16, 17, or 18).
3. Waves included 2011–2013, 2013–2015, 2015–2017, and 2017–2019.
4. The NSFG did not ask women attitudes about masculinity (nor femininity). Thus, although women were sampled for the NSFG, this paper does not analyse women’s attitudes because they were not asked the questions about masculinity.
5. The Cronbach’s alpha for the three attitudes was .51. The Cronbach’s alpha for the attitude about showing pain and going to the doctor – which are, presumably, more related – was similarly low: .56. Thus, none of the attitudes could be combined.
6. The four-option version of attitudes could not be analysed using ordered logistic regression because they violated the proportional odds assumption, as did the five-option version of attitudes. Ordered logit models are thus statistically inappropriate.
7. Few substantive differences emerged in robustness tests that categorized neutral and ‘don’t know’ responses with conservative responses (when these were coded as 1 instead of 0 when dichotomized), since there were so few. The frequency of the two combined was 0.94% on the attitude about showing pain, 0.47% on the attitude about going to the doctor, and 4.56% on the attitude about men’s sexual needs.
8. A limitation of this measure is that individuals who immigrated as children would be classified similarly as individuals who immigrated as adolescents or adults.
9. The NSFG did not provide detail about respondents’ country of origin.
10. Respondents were classified as a sexual minority if they identified as bisexual, gay, ‘something else,’ or ‘don’t know’ and reported at least one lifetime male sexual partner. Respondents who reported one of those four identities but did not report at least one same-sex partner were excluded from analyses. This design was necessary because the prevalence of sexual minority identification is very small compared to heterosexual identification, so errors in sexual minority status (if respondents misunderstood the question about sexual identification or mistakenly answered it) would greatly affect estimates.
11. The margins command was combined with options for post, subpop, and vce(unconditional).
12. Differences are calculated using the lincom command.