Implicit racism, colour blindness, and narrow deﬁnitions of discrimination: Why some White people prefer ‘All Lives Matter’ to ‘Black Lives Matter’

The Black Lives Matter (BLM) movement has been called the ‘civil rights issue of our time’ (Holt & Sweitzer, 2020, Self and Identity , 19(, p. 16) but the All Lives Matter (ALM) movement swiftly emerged as an oppositional response to BLM. Prior research has investigated some predictors of support for ALM over BLM, but these predictors have thus far not included levels of racial bias or potentially relevant constructions of racism. This pre-registered, cross-sectional study ( N = 287) tested the degree to which White participants’ support for ALM could be predicted using measures of racism (implicit and explicit) and ideological stances around the construction of ‘racism’ (that discourage the recognition of contemporary inequalities and discrimination). Using multiple regression analyses, we found that implicit racism, colour-blind ideology, and narrow deﬁnitional boundaries of discrimination positively predicted support for ALM over BLM. Explicit racism, collective narcissism, and right-wing political orientation did not predict ALM support, nor did any (2-way) interaction of these predictors. Implications for our understanding of the All Lives Matter movement are discussed.

The Black Lives Matter (BLM) movement has been called the 'civil rights issue of our time' (Holt & Sweitzer, 2020, Self and Identity, 19(, p. 16) but the All Lives Matter (ALM) movement swiftly emerged as an oppositional response to BLM. Prior research has investigated some predictors of support for ALM over BLM, but these predictors have thus far not included levels of racial bias or potentially relevant constructions of racism. This pre-registered, cross-sectional study (N = 287) tested the degree to which White participants' support for ALM could be predicted using measures of racism (implicit and explicit) and ideological stances around the construction of 'racism' (that discourage the recognition of contemporary inequalities and discrimination). Using multiple regression analyses, we found that implicit racism, colour-blind ideology, and narrow definitional boundaries of discrimination positively predicted support for ALM over BLM. Explicit racism, collective narcissism, and right-wing political orientation did not predict ALM support, nor did any (2-way) interaction of these predictors. Implications for our understanding of the All Lives Matter movement are discussed.
Racism affects almost every aspect of contemporary life (Pearson, Dovidio, & Gaertner, 2009;Tatum, 1999;Williams, 2019). Even when factors other than race are equivalent or accounted for, research shows that ethnic minorities are offered less support in higher education (Milkman, Akinola, & Chugh, 2015), are less likely to receive offers of employment (Bertrand & Mullainathan, 2004;Booth, Leigh, & Varganova, 2012;Eaton, Saunders, Jacobson, & West, 2020), are treated less politely and offered less help in stores (Bourabain & Verhaeghe, 2018), are treated with more suspicion in public places (Schreer, Smith, & Thomas, 2009), are considered less desirable as romantic partners (Mendelsohn, Shaw Taylor, Fiore, & Cheshire, 2014;West, 2019a;West, Lowe, & Marsden, 2017), are judged more harshly for the same behaviours (Sommers & Ellsworth, This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. *Correspondence should be addressed to Keon West, Department of Psychology, Goldsmiths, University of London, London, SE14 6NW, UK (email: keon.west@gold.ac.uk).
A noteworthy point is that racism is not merely widespread among individuals, but part of the structure of contemporary society (Salter, Adams, & Perez, 2018). That is, at the simplest level, racism may be understood as (demonstrable) differences in responses to individuals based on race (Dovidio, Kawakami, & Gaertner, 2002;West, 2019b), in line with broader individualist Western ideologies (Adams, Edkins, Lacka, Pickett, & Cheryan, 2008;Henrich, Heine, & Norenzayan, 2010;. However, at a more nuanced and complete level, racism may be understood as intersecting individual, 'institutional, systemic, and cultural practices that perpetuate and maintain race-based hierarchies' (Salter et al., 2018, p. 151).
It is thus unsurprising that racism also manifests in policing. For example, the United Kingdom's (UK's) 'stop and search' laws (Bowling & Phillips, 2007, p. 936) have been disproportionately used to target and harass Black people, particularly Black men. Depending on the region and the year, Black people in the UK were between 6 and 27 times more likely to be stopped and searched by the police than their White counterparts (Bowling & Phillips, 2007;Joseph-Salisbury, Connelly, & Wangari-Jones, 2020). Similarly, large-scale analyses of police stops in the United States showed that Black drivers are more than twice as likely to be stopped than White drivers (Pierson et al., 2020). Furthermore, when they do encounter the police, 'Black and White civilians experience fundamentally different interactions' (Kramer & Remster, 2018, p. 960), particularly as it concerns the use of force (Kahn, Goff, Katherine Lee, & Motamed, 2016).
The use of force is an important consideration. Police officers in the United States shoot and kill approximately one thousand people each year (Statista, 2021a). Experiments using computer simulations have shown that police officers (and civilians) are more likely to erroneously shoot an unarmed Black person than an unarmed White person and conversely fail to shoot an armed White person than an armed Black person (Correll, Park, Judd, & Wittenbrink, 2002, 2007Plant & Peruche, 2005). It is therefore not surprising that the rate of fatal shootings (per capita) by the police is approximately 2.5-3 times higher for Black Americans than for White Americans (Edwards, Lee, & Esposito, 2019;Schwartz & Jahn, 2020;Statista, 2021b). These differences persist even when other relevant factors (e.g., civilian behaviour, socio-economic status) are accounted for. Indeed, research shows that Black victims of police shooting are twice as likely to be unarmed as White victims (Nix, Campbell, Byers, & Alpert, 2017).
The Black Lives Matter movement (BLM) arose in response to these race-based discrepancies in treatment by the police, as well as to the apparent impunity with which police and White civilians use force (even deadly force) against Black civilians (Holt & Sweitzer, 2020). Though the movement is decentralized, it started in the United States in 2012 after George Zimmerman (Hispanic, Caucasian, and armed with a gun) employed the 'stand your ground' legal defence (Holt & Sweitzer, 2020, p. 16) and was acquitted of killing Trayvon Martin (17 years old, unarmed, African American). The movement gained further traction following the killings of Eric Garner, Michael Brown, Tamir Rice, Freddie Gray, Sandra Bland, Philando Castile, Stephon Clark, Breonna Taylor, and George Floyd (Giorgi et al., 2020). All were African American, unarmed, and killed by police officers who later faced few or no consequences (at the time of writing, Derek Chauvin's sentence for the murder of George Floyd had not yet been decided).
Though police officers in the United Kingdom kill far fewer people than their counterparts in the United States, there are nonetheless similar racial disparities in the treatment, use of force, and use of deadly force against Black British people compared with White British people (Joseph-Salisbury et al., 2020). To illustrate, in May of 2020 the Torquay police arrested Simeon Francis (Black, British, 35 years old) who later died in their custody in unexplained circumstances (BBC, 2020). A video of an earlier arrest shows the police pushing Mr Francis against a wall, dragging him to the ground, and (reminiscent of the killing of George Floyd) putting a knee on his back while he said 'I can't breathe' (BBC, 2020). In another example, police in Tottenham shot Millard Scott (Black, British, 62 years old) with a taser in his own home (Taylor, 2020). The police shot the elderly man while he was at the top of the stairs, which further endangered his life as he lost consciousness and fell down the stairs. This experience is not unique, with recent statistics showing the police are nice times more likely to draw tasers on Black British people than on White British people (Mellor, 2021 (Joseph-Salisbury et al., 2020). As in the United States, the police officers involved in these incidents often faced few or no consequences. Unsurprisingly then, the BLM movement has become a British phenomenon as well as an American one, and one that is echoed in many other countries as well.
The All Lives Matter movement (ALM) emerged in response to and as a criticism of the Black Lives Matter movement (Giorgi et al., 2020). However, the motivations behind the ALM movement are a subject of significant debate. Supporters of ALM claim that their colour-blind position is more inclusive, seemingly interpreting 'Black Lives Matter' as 'Only Black Lives Matter' (Atkins, 2019). Former President Donald Trump and several other high-level Republican representatives have described BLM as racist and divisive, and expressed a clear preference for terms like All Lives Matter (Weigel, 2016). High-profile Democrats, including past presidential candidate Hillary Clinton, have also used the phrase 'All Lives Matter' in apparent attempts to promote inclusivity (Rappeport, 2015). However, others see these responses to BLM as disingenuous and suggest that the promotion of ALM reveals much more negative motivations than this apparent inclusivity (Lebron, 2017).
Very little empirical research has investigated the motivations behind the rejection of BLM and support for ALM. Hayward, Hornsey, Tropp, and Barlow (2017) found that negative contact with Black people made White people less likely to blame a White officer when a Black man died in his custody. Kilgo and Mourão (2019) found that consuming conservative media was associated with more negativity towards BLM, though liberal media was not associated with positivity towards BLM. Holt and Sweitzer (2020) found that higher levels of social dominance orientation (a preference for social hierarchies, rather than equality) predicted more support for ALM. However, research to date has curiously ignored both racism and constructions of racism when considering potential predictors of support for ALM. This is a significant gap which we aim to address in this current research.
Possible predictors of support for all lives matter Implicit and/or explicit racism In some ways, racial prejudice seems like the most parsimonious explanation of negativity towards BLM and support for ALM. BLM is an explicitly anti-racist movement (Atkins, 2019;Lebron, 2017). 'It takes no advantages from White [people]. It just asks that [Black people] be treated with the same caution and respect as White [people] by police' (Holt & Sweitzer, 2020, p. 16). Those who prefer the current racial hierarchy in which White people receive superior treatment would likely oppose any efforts towards racial equality. Claims that opposition to BLM are based on inclusivity seem undermined by the promotion of Blue Lives Matter as a response to BLM, as Blue Lives Matter (at the very least) shares the same problems of exclusivity (Giorgi et al., 2020;Solomon, Kaplan, & Hancock, 2019).
Moreover, criticism of BLM is frequently paired with racist sentiments. For example, Councillor David Jack of Sandbach, Cheshire aggressively opposed the Black Lives Matter protests organized in that town (Campbell & Mohdin, 2020). He was also an administrator of a Facebook group (Sandbach Sarcastic Society) in which private information about the BLM protestors was published, alongside comments that 'All Lives Matter', and threats to send the organizers' addresses to the English Defence League and the Klu Klux Klan (Campbell & Mohdin, 2020). Former New York mayor Rudolph Giuliani said 'I don't see what Black Lives Matter is doing for Blacks other than isolating them. All it cares about is the police shooting of Blacks. It doesn't care about the 90 percent of Blacks that are killed by other Blacks' (Weigel, 2016, p. 1).
An interesting question then is whether explicit or implicit racism would better predict support for ALM over BLM. On the one hand, despite the pervasiveness of contemporary racism, social norms in Western societies still strongly prohibit the expression of overtly racist sentiments (Crandall, Eshleman, & O'Brien, 2002;Gawronski, Peters, Brochu, & Strack, 2008;Greenland et al., 2018;West & Hewstone, 2012). Those who explicitly hold racist beliefs may thus be unwilling to express them, but these explicit racist beliefs may nonetheless predict their responses to other social events such as BLM (White & Crandall, 2017). On the other hand, implicit bias is also a widespread concern (Nosek, Greenwald, & Banaji, 2007), and it is plausible that those who oppose BLM may be unaware of the racial bias motivating that opposition. Thus, it is important to consider both implicit and explicit racism as predictors of support for ALM.

Colour blindness
Beyond racial bias itself, certain ideological stances concerning racism may explain discomfort with BLM and a preference for ALM. ALM is explicitly colour-blind; that is, the movement endorses 'the belief that race should not and does not matter' (Neville, Lilly, Lee, Duran, & Browne, 2000, p. 60; see also Bonilla-Silva & Forman, 2000). While this may initially seem like an egalitarian stance, it also frequently reflects a limited understanding of the ways bias and discrimination operate and affect Black people and other minorities, as well as an unwillingness to acknowledge racism in contemporary society or to do anything to address that racism (Tawa, Ma, & Katsumoto, 2016). Consequently, research finds that colour-blind ideology leads to more racial bias, not less, and that colour blindness can be employed as a way to downplay or ignore contemporary inequalities (Leslie, Bono, Kim, & Beaver, 2019;Richeson & Nussbaum, 2004). Colour-blind White people appear less friendly to Black interaction partners (Norton, Sommers, Apfelbaum, Pura, & Ariely, 2006) and are less likely to recognize racial discrimination when it occurs (Offermann et al., 2014). Indeed, colour-blind ideology is better than explicit modern racism at predicting negativity towards actions that address racism (e.g., affirmative action; Awad, Cokley, & Ravitch, 2005). A commitment to colour blindness may then explain why some people find the language and goals of BLM to be uncomfortable and confrontational; they undermine a colour-blind individual's ability to ignore the impact of race and racism in contemporary society (Atkins, 2019;Tawa et al., 2016;Weigel, 2016).
Where people draw this metaphorical line between 'discrimination' and 'not discrimination' can be referred to as their Definitional Boundaries of Discrimination (i.e., DBDs). These definitional boundaries can be very broad (i.e., including a range of different acts up to the most subtle) or very narrow (i.e., excluding everything but the most overt acts). Using very broad boundaries of discrimination makes it more likely that an individual would perceive discrimination in contemporary society, while very narrow boundaries would make an individual unlikely to perceive discrimination (Greenland, van Laar & West, under review). BLM was founded as a response to contemporary discrimination in the use of force against Black people (Lebron, 2017). Individuals who do not recognize contemporary discrimination are thus probably less likely to perceive the need for a movement like BLM and would likely prefer the less specific message of ALM.

Collective narcissism
De Zavala, Cichocka, Eidelson, and Jayawickreme (2009) proposed the concept of collective narcissism: a 'narcissistic ingroup positivity' (Cichocka, 2016), characterized by an grandiose image of one's social group, a strong desire to protect and bolster the image of that group, and a reliance on external validation. Collective narcissism resembles individual narcissism in that it presents an inflated self-love and intense defensiveness when this selflove is not validated by external recognition (Cichocka, 2016;de Zavala & Cichocka, 2012;de Zavala et al., 2009). Prior research has shown that individuals high in collective narcissism are very resistant to negative messages about their group (Cichocka, 2016;de Zavala et al., 2009). Thus, a White person high in collective narcissism may respond negatively to BLM's assertions about the pervasive racial discrimination and the impunity of police officers. Suggestions that contemporary (White) society is still pervasively racist may be perceived as 'heated and accusatory' (Weigel, 2016, p. 2;DiAngelo & Sensoy, 2012), leading to negative attitudes towards BLM and a preference for ALM.

Political affiliation
There is evidence that (in the United States at least) political affiliations are strongly related to judgements about racism. For example, Dost, Enos, and Hochschild (2020) found that Democrat loyalists were much more likely to describe events (including the police shootings of civilians) as racist compared to Republican loyalists. Accusations and counter-accusation of racism, calls for free speech, and consequent claims of victimization are a key element in mobilizing support for right-wing populist movements (Durrheim et al., 2018). It is therefore possible that preferences for ALM are largely due to political identity, rather than racial bias or ideologies concerning racism per se. In order to demonstrate the importance of racism and racism-related ideologies, this research must show that they predict responses to ALM and BLM even when political affiliation is accounted for.

Current research
This cross-sectional study investigated the extent to which White British participants' preference for the statement 'All Lives Matter' over 'Black Lives Matter' can be predicted with anti-Black racism (both implicit and explicit), factors that discourage the recognition of contemporary anti-Black racism (colour blindness, narrow definitional boundaries of discrimination, and collective narcissism), and political affiliation (left wing vs right wing). Though we expected all the aforementioned predictors to predict a preference for 'All Lives Matter', we remained agnostic concerning the variables that would explain the most variance. We also investigated whether preferences for 'All Lives Matter' could be predicted by any two-way interactions between the aforementioned predictors. All methods and hypotheses were pre-registered (https://aspredicted.org/blind.php?x= wz2zn5).

Participants and procedure
To determine the sample size necessary for this study, we conducted a-priori power analyses with G*Power (Faul, Erdfelder, Buchner, & Lang, 2009;Faul, Erdfelder, Lang, & Buchner, 2007). Assuming a multiple regression analysis, a medium effect size, that is, f 2 = .15, a = .05, b = .8, and 28 predictors (to account for each of the six independent variables and all possible two-way interactions between independent variables), we found that 181 participants would be sufficient for adequate power. Anticipating some attrition, we set the upper limit on the number of participants recruited to 300.
Hence, using a combination of online research panels (via Prolific) and word-of-mouth recruitment by a team of three research assistants, 300 participants were initially recruited. All participants were White, British, and living in the United Kingdom. Of those participants, 13 withdrew almost immediately and provided no useful data. Of the remaining 287 participants, 18 had missing data for at least one of the variables of interest. This accounts for the fluctuations in degrees of freedom in the analyses below. Nonetheless, we retained all these participants in order to provide the maximum power possible for each analysis. Of those 287 participants (mean age = 35.56, SD = 15.35), 95 (33.1%) identified themselves as 'men', 181 (63.1%) identified themselves as 'women', 2 (.7%) identified themselves as 'other', and 9 did not identify a gender. Of those who reported a level of education, we noted that for 8.7% of the participants, their highest level of education was GCSE or equivalent; for 41.2%, it was A-levels or equivalent; for 33.2%, it was a BA or equivalent; for 15.5%, it was an MA or equivalent; and for 1.4%, it was a PhD or equivalent. Thus, though education was not a variable of interest, we noted that our sample was reasonably representative of the United Kingdom; 50.1% of our participants had a tertiary degree, while 52% of 25-34 year-olds in the United Kingdom have a tertiary degree (OECD, 2020).
Participants were told that the study was investigating contemporary racial issues, but were not made aware of the specific hypotheses or variables of interest. All participants completed the measures using a computer, and all variables were presented in an order that was randomized for each participant. After completing the measures of implicit racism, explicit racism, colour blindness, definitional boundaries of discrimination, collective narcissism, and political affiliation, participants were thanked, fully debriefed, and offered pay equivalent to £7 per hour.

Measures
All lives matter Preference for 'All Lives Matter' over 'Black Lives Matter' was assessed with a 6-item scale developed for this study. On 100-point sliding scales anchored by two extremes (1 = Black Lives Matter and 100 = All Lives Matter), participants responded to six questions (e.g., 'Which of these two statements would you be more comfortable saying in public?'; see Appendix 1). Principal components analysis of the six items revealed a single factor onto which all items loaded well (.86 < k < .97), and Cronbach's alpha scores showed the scale to be highly reliable (a = .97).

Implicit racism
Participants completed a version of the Black-White Implicit Association Test or IAT (Greenwald, McGhee, & Schwartz, 1998) programmed by IATGEN (Carpenter et al., 2021). Despite some limitations, the IAT is still widely used and widely regarded as one of the best measures of implicit bias (Carpenter et al., 2021;Nosek et al., 2007;Project Implicit, 2018). Like the original Black-White IAT, this version assessed the strengths of participants' associations between the categories Black and White and the attributes of good and bad. The design and analysis of this IAT was the same as that of the Black-White IAT used by Project Implicit (Project Implicit, 2018). We used the same picture stimuli and the same word stimuli.
As in the original, our IAT consisted of seven phases, some of which were practice tasks to acquaint participants with the stimuli and sorting rules. In the critical phases, participants were asked to simultaneously sort stimuli and either (1) use the same response keys for Black and good and for White and bad or (2) use the same response keys for White and good and for Black and bad. Implicit bias scores were calculated by the IATGEN programme using the same algorithms used in previous IAT studies (Greenwald & Nosek, 2003). To simplify, this was done by comparing the speed of responding and the frequency of errors in the critical phases (i.e., comparing the strength of the association between Black and good, and between White and good). The notable difference (and advantage) in the IATGEN version is the ability to use it online and without the requirement of third-party tools (Carpenter et al., 2021). Higher values indicated higher levels of implicit anti-Black racism.

Explicit racism
Participants completed a version of the Modern Racism Scale (McConahay, 1986;Morrison & Kiss, 2017) referring to the United Kingdom rather than the United States where necessary (a = .84). They responded to seven items (e.g., "Blacks should not push themselves where they are not wanted" and "Blacks have more influence on society than they ought to have") using a 100-point sliding scale anchored at two extremes (Strongly Disagree and Strongly Agree). Higher values indicated more explicit racism.

Colour blindness
Participants completed a version of the 20-item Colour-blind Racial Attitudes Scale (Neville et al., 2000) referring to the United Kingdom rather than the United States where necessary (a = .89). They responded to all items using a 6-point Likert scale (1 = Strongly Disagree, 6 = Strongly Agree; e.g., It is important that people begin to think of themselves as British and not Black British or Asian British). Higher values indicated a preference for colour blindness.

Definitional boundaries of discrimination
Participants completed the 15-item (a = .85) DBD measure (Greenland, van Laar & West, under review). They responded to all items using a 7-point Likert scale (1 = Strongly Disagree, 7 = Strongly Agree). Example items include the following: 'The core of anti-Black racism is that it is malicious: if a person is not being malicious, then it can't be racism'. Higher values indicated more narrow definitional boundaries of discrimination, that is, a tendency to see only the most overt acts as discriminatory.

Collective narcissism
Participants completed the 9-item (a = .92) Collective Narcissism Scale (de Zavala et al., 2009). They responded to all items using a 6-point Likert scale (1 = Strongly Disagree, 6 = Strongly Agree). Example items include the following: 'I will never be satisfied until my group gets the recognition it deserves'. Higher values indicated higher levels of collective narcissism.

Political affiliation
This was measured by a single item in which we asked participants, 'How would you describe yourself politically?'. Participants responded using a 100-point sliding scale anchored at Extremely Left Wing and Extremely Right Wing. Higher values indicated a more right-wing political position.

Preliminary results
Descriptive statistics and correlations between variables are shown in Table 1. All proposed predictors were significantly correlated with support for ALM (all ps < .001). Most predictor variables were also significantly correlated with all other predictor variables; the exception was collective narcissism, which was not correlated with implicit racism, colour blindness, or definitional boundaries of discrimination. We noted that mean values of all variables differed significantly from the mid-points of the scales (all p < .001). Implicit racism scores were higher than the mid-point of the scale, but all other scores were lower than the mid-point of the scale. In other words, this indicates that our participants showed significant implicit anti-Black bias, but were otherwise predominantly left wing, and low in explicit racism, colour blindness, (narrow) definitional boundaries of discrimination, and collective narcissism.
In line with our pre-registration, we also conducted a multiple regression analysis in which all six predictor variables and all possible 2-way products of the predictor variables (e.g., implicit 9 explicit racism, implicit racism 9 colour blindness, implicit racism 9 DBDs, etc.) were entered simultaneously as predictors. Though the overall model was again significant F (21, 263) = 17.59, p < .001, R 2 = .60, none of the independent variables or products predicted support for ALM (À.52 < all bs < .72; .06 < all ps < .97). Furthermore, the addition of the products as predictors only slightly increased the explained variance in support for ALM: R 2 = .60 vs. R 2 = .58.
Thus, we found no evidence that any interaction of the proposed variables predicted support for ALM. However, we did find evidence that implicit racism, colour blindness, and narrow definitional boundaries of discrimination (against Black people) all positively predicted support for ALM over BLM.

Discussion
The study was designed to test the extent to which White participants' support for 'All Lives Matter' (compared to 'Black Lives Matter') could be explained by variables associated with anti-Black racism and/or variables that discourage the recognition of contemporary anti-Black racism. The results indicated a significant association between ALM support and implicit racism, colour blindness, and narrow definitional boundaries of discrimination (DBDs). The more participants said that they supported ALM, the more they (1) showed high levels of implicit racism against Black people; (2) endorsed colourblind ideologies that downplayed or ignored contemporary inequalities (Leslie et al., 2019;Richeson & Nussbaum, 2004); and (3) defined racism in such narrow terms as to make them unlikely to see racism except in the most egregious circumstances (Greenland et al., 2018(Greenland et al., , 2019. Importantly, these variables are all either direct measures of anti-Black racism (Greenwald et al., 1998) or have been shown to be strongly associated with anti-Black racism (e.g., Greenland, van Laar & West, under review;Leslie et al., 2019;Richeson & Nussbaum, 2004). These variables predicted support for All Lives Matter over Black Lives Matter even when political orientation was accounted for (i.e., included as a simultaneous predictor in a multiple regression analysis, rather than merely a correlation analysis). Thus, although supporters of ALM argue that it is more inclusive, they seem to be motivated by anti-Black racism and factors associated with the downplaying of anti-Black racism.
At the risk of falling victim to hindsight bias, one could argue that these results are not particularly surprising. We acknowledge that supporters of ALM have explicitly denied being motivated by racism and have indeed attempted to position themselves as the less racist movement (Tawa et al., 2016). Nonetheless, it seems apparent that the people who prefer and/or benefit from a racialized status quo would also be the people who oppose any challenges to that racialized status quo (Salter et al., 2018). Regardless of the predictability of the results, we must note the curious disinterest shown by empirical research to date concerning the relationship between racism and support for the All Lives Matter movement. While research looking at contact and the media as predictors of ALM support are valuable (Hayward et al., 2017;Holt & Sweitzer, 2020;Kilgo & Mourão, 2019), the total lack of research on racism as an explanation for All Lives Matter support seems itself to be an example of collective colour blindness: avoiding the direct consideration of racism as an explanation for behaviour, even when it seems to be the most evident and parsimonious explanation.
In any case, it is useful and important to provide concrete evidence that supports what many have argued about ALM: that the motivations of its supporters include implicit racism. Participants' support for ALM seems to be driven by the same processes that brought about the BLM movement in the first place: a refusal to engage with the reality of contemporary forms of anti-Black racism; an insistence that acknowledging the experiences of Black people is part of the problem; and a creeping anti-Black bias that derives from a racialized social order. These provide the conditions in which racism can be acted out with impunity: Actors are not only motivated to deny anti-Black racism, but they have the tools (i.e., colour-blind ideologies and narrow definitional boundaries of discrimination) with which to do it.
The current study benefits from a relatively diverse, non-student sample, a-priori power analyses, and pre-registration. These strengths add to confidence in and generalizability of the findings (Faul et al., 2009;Henrich et al., 2010). Nevertheless, the data are correlational: We have not demonstrated that anti-Black racism causes support for ALM. It is possible that there are third variables that are driving both effects (e.g., social dominance orientation, a lack of empathy, or threat) or even that participation in ALM changed or consolidated participants' racial attitudes (i.e., reverse causation; Drury & Reicher, 2000). It would be worth replicating the study with additional variables and/ or extending it to show that the ALM argument is only deployed in contexts where anti-Black racism is implicated (e.g., O'Brien, Garcia, Crandall, & Kordys, 2010;White & Crandall, 2017). Other future research might go beyond individual-level variables, considering factors such as social norms, conceptualizations of Whiteness, and institutional responses to BLM and ALM.
One might also argue that the design constructed a false dichotomy between BLM and ALM: the dependent variable placed ALM and BLM as opposite extremes, when it is possible that participants might like and appreciate both statements (or indeed dislike both). While we acknowledge this limitation, we still think this research was a necessary first step given the social and political context. ALM arose as a direct response to BLM (Giorgi et al., 2020) and whatever the two statements might mean in a different context, they meant something specific here. It was this meaning that interested us the most. Also, for each item participants were free to choose the middle point of the response scale, indicating they agreed with both (or neither) ALM and BLM. The fact that participants generally did not choose this option and that we found reliable relationships between ALM support and all predictor variables (see Table 1) suggests that the participants' choice of ALM vs BLM was indeed related to anti-Black racism and associated factors.
What does this research tell us about achieving social change? On one hand, it demonstrates (again) the difficulty of achieving change, particularly when that change can be constructed as a challenge to the status quo and associated belief systems (Knowles, Lowery, & Schaumberg, 2014). Our analysis of the relation between ALM and implicit racism is unlikely to persuade those who identify as ethno-nationalists (and indeed could be used rhetorically to shore up further support for right-wing populist activity; Durrheim et al., 2018). It might, however, persuade those who are 'BLM-hesitant': those who consider themselves to be political centrists but who are nevertheless uncomfortable with the direct stance of BLM. We note that this description could apply reasonably well to our participant sample, which was predominantly left wing and low in explicit expressions of anti-Black racism, but nonetheless showed significant levels of implicit anti-Black racism. Talking frankly about the causes and consequences of colour blindness and definitional boundaries of discrimination might be a place to start with such individuals. There is also reason to be optimistic in that the BLM movement itself might be consequential in reducing implicit racism (Sawyer & Gampa, 2018).

Conclusions
A pre-registered cross-sectional study found that anti-Black racism predicts White participants' support for All Lives Matter, even when controlling for political affiliations. The results support the suggestion that preference for All Lives Matter over Black Lives Matter is indicative of both implicit racism and ideological stances that minimize or discourage the recognition of contemporary forms of racial discrimination. We suggest that constructions of 'racism' that enable participants to minimize and resist claims of discrimination (i.e., colour-blind ideology and narrow definitional boundaries of discrimination), thereby (ironically) perpetuate racism and impede movements (such as Black Lives Matter) intended to reduce racism.