Political Protesting, Race, and College Athletics 1015
respondent reported being an African American.
18
We included three types of control vari-
ables: (1) demographics including gender, age, income, and education; (2) political/social
attitudes including ideology, concern about police conduct, American identity, skepticism
about racial discrimination, and political interest; and (3) career characteristics including
playing a sport in college, coaching men, years in the field, head coach status, Division III
status, and coaching basketball or football (as these two sports stand out in terms of revenue
generation). We suspect conservatives and those with greater discrimination skepticism to
be less supportive of protests of any kind, as protests seek to change the status quo and tend
to be racialized (see Druckman, Howat, and Rodheim, 2016; Quinnipiac University Poll
2016). Police conduct concern and American identity may matter for the national anthem
and political apparel protests given their connections to police behavior and, potentially,
patriotism. We do not include these two variables in the APU models since that protest
is orthogonal to concerns about the police or patriotism (including the variables does not
change the results).
We present the results for the national anthem, APU, and political apparel protests, re-
spectively, in Tables 1–3.
19
We find that race is statistically significant and has a substantively
large effect in every model. African-American coaches, relative to non-African-American
coaches, display vastly more support for all three types of protest and clearly oppose team
rules that would disallow such protests. To gauge the substantive effect of race, we calculate
the predicted mean scores for non-African Americans and African Americans, holding all
other variables constant at their means (using Clarify; see King, Tomz, and Wittenberg,
2000). Recall, both the support and the team rule variables are five-point scales. African-
American coaches exhibit nearly a full point greater average support score (from 2.26
[SE = 0.05] to 3.17 [SE = 0.19]) for the national anthem protest, and are roughly a half
point (from 3.09 [SE = 0.06] to 2.62 [SE = 0.23]) more opposed to a team rule. We find
similar results for the APU and political apparel protests, albeit slightly smaller differences
in support. Specifically, support for APU protests increases from 3.04 (SE = 0.05) among
non-African Americans to 3.69 (SE = 0.18) among African Americans, and support for a
team rule against them decreases from 2.49 (SE = 0.05) to 1.88 (SE = 0.18). Support for
apparel protests increases from 2.64 (SE = 0.04) to 3.25 (SE = 0.16), a nd support for a
team r u le decreases from 2.90 (SE = 0.06) to 2.18 (SE = 0.21).
20
The race gap echoes the aforementioned divides in other populations (e.g., the public,
student athletes). Of equal, if not greater, importance are results on attributions for the
protests.
21
Across all three protests, we find African-American coaches, relative to non-
African-American coaches, are much more likely to believe the student athletes who protest
do so because they care about the issue. They report much lower scores on beliefs that the
protests stem from social pressure or an effort to garner personal attention. We calculate
the probability (holding all other variables at their mean scores) of believing the given
cause is why the student athletes protest. For the national anthem protest, there is a 0.92
18
We do not include variables for other races/ethnicities since our expectations revolve specifically around
African Americans rather than minorities in general. Practically, the sample also includes very few other
minorities (e.g., just 3 percent of the sample reported being Hispanic).
19
The Ns for the regressions shrink for two reasons. First, of the 873 coaches who started the survey, a fair
number rolled off, with 662 answering at least one of the three overall protest support questions. Second, there
was some nonresponse on three items: racial discrimination skepticism, American identity, and police conduct
concern batteries. The central results are robust if we exclude these variables.
20
While these race effects are quite notable, support even for African-American coaches is not extremely
high, never reaching an average score of 4 on the five-point support scale. Across the entire sample, support is
fairly low: 2.33 for the national anthem protest, 3.09 for the APU protest, and 2.68 for the political apparel
protest.
21
Recall that respondents could agree with multiple attributions.