Informing Just Design with Place-Based Racial History

In How Racism Takes Place, Lipsitz (2011) writes that “by examining residential and school segregation, mortgage and insurance redlining, taxation and transportation politics, or the location of environmental amenities and toxic hazards, we learn that race is produced by space, that it takes places for racism to take place”. Race and the built environment in the United States are inextricably linked because of explicit decisions made in history that continue to affect the present (Lipsitz 2011; Rothstein 2017). This, in addition to the gap between racial groups in perceptions of contemporary racism (Norton and Sommers 2011; Strickhouser et al. 2019), were motivation for the study of place-based critical history in the field of psychology. Place-based critical history is a narrative of history presented with the perspective of the present in mind (Bonam et al. 2019). Empirical work has used place-based critical racial history as a tool to increase the recognition of contemporary racism in the United States (Bonam et al. 2019; Coleman et al. 2019). This work builds on the Marley hypothesis literature, which posits that the relevance of racial identity and a knowledge of racial history are two factors that help explain the gap in perceptions of contemporary racism (Nelson et al. 2013). There are boundary conditions to the Marley hypothesis: if perceptions are being assessed at a limited number of institutional or geographic regions, the effects noted previously have not always been successfully replicated (Bonam et al. 2019; Strickhouser et al. 2019). Specifically, findings produced at separate academic institutions in the Midwest did not replicate to a single more racially diverse academic institution in the southeast United States. This is a factor that specific institutions need to be aware of when looking to apply the findings of the Marley hypothesis.

This study seeks to expand the work using place-based critical racial history as an intervention by focusing on how it may affect participants’ reactions to a climate adaptation design decision context. By expanding this work to focus on design, the operationalization of this concept will need to change. Specifically, the intervention will need to be much shorter to feasibly fit into a presentation on a climate adaptation project, where it is one of many factors competing for attention.

There are many types of justice that may work synergistically with a knowledge of place-based critical racial history to improve equity in the built environment. For example, Bozeman et al. (2022) explore how the complexities of resource, policy, and cultural needs overlap to move toward or away from systemic equity, which incorporates different aspects of distributive, procedural, and recognition justice. There is also restorative and intergenerational justice, which are highly relevant to climate conversations (Siders et al. 2021). These are promising areas for future work. Inspired by Lipsitz’s observation of the relationship between race and space, this study focuses on distributive justice, the allocation of benefits and burdens.

There are at least four principles of distributive justice that describe common ways in which people describe allocations of resources as fair or just; an individual will likely prefer different principles depending on the context (Deutsch 1975; Konow 2001). The principle of equality suggests that resources should be allocated equally among a population, the principle of equity suggests that resources should be allocated with regard to individual contributions, the principle of need suggests that resources should be allocated to prioritize those in need, and the principle of entitlement suggests that resources should be allocated by status (Gollwitzer and van Prooijen 2016). Equality implies that frontline communities will get an equal portion of resources, presumably less than they need to be successful in adapting to climate change because of the historical and complex factors that have systemically disadvantaged them. Next, it should be noted that the term equity has many definitions that are very different from each other (Pereira et al. 2017). Equity has been defined to mean impartial (Sen 2009). It has also been defined as broadly as distribution based on individual differences (Rawls 1999). By some, it is understood as equivalence of the outcome/input ratios, which is the concept that this study refers to as merit (Cook and Hegtvedt 1983). Merit is likely to be a direct reinforcer funneling attention and resources away from frontline communities that do not have as much financial capital to invest or as much temporal wealth to invest. However, merit framed in other ways may work to the benefit of frontline communities. For example, if a metric considered in distributive decision-making was focused on carbon footprint, frontline communities may fare better than communities with individuals who have a lot of disposable income to spend on consumeristic priorities. Lastly, need-based allocation strategies emphasize attention to those who are working at a disadvantage and what they need to be successful. What success means if defined quite differently based on the various strategies. For example, sufficientarianism, which focuses on whether individuals have enough of a resource, might be a less effective position for supporting frontline communities because what is needed as a bare minimum will be just as unpredictable and rapidly changing as the effects of climate change.

Equality and equity (i.e., merit) have received much more empirical attention than need and entitlement, both of which have been tied to the environmental justice movement (Gabriel and Bond 2019). It is possible to apply multiple principles in the same distributive justice effort. However, equality and need are often confounded, which makes them difficult to study together in an empirical manner (Hülle et al. 2018).

Need, particularly, has potential to be of help to frontline designers because it is “more fundamental than the other principles” in that it “refers to the basic conditions for human existence” (Traub and Kittel 2020). The question of what constitutes need has been a subject of debate. In a built environment context, clean air and water would almost certainly be included as needs, but is green space also included in this definition, given the empirical evidence on how it can impact well-being? (Clayton et al. 2016). There is no definitive answer to this question, and given the unique context and composition of each frontline community, it is perhaps more beneficial for these questions to be left to the discretion of frontline designers and professional designers in each context.

A popular theory associated with need-based distributive justice and climate adaptation is the Capabilities Approach, which emerged from a body of work by Sen and Nussbaum (Nussbaum 1988; Sen 2009; Traub and Kittel 2020). The Capabilities Approach is a framework that emphasizes the importance of well-being, which is defined by capabilities and functionings. Capabilities describe what is feasible and functions are capabilities that have been actualized. This work has been applied throughout the discourse on climate adaptation. For example, it was discussed as a method for articulating individual climate adaptation enablers and barriers to well-being (Kronlid 2014). It has also been lauded as offering “a particularly constructive way of understanding issues of vulnerability and impact” in a climate adaptation context (Schlosberg 2012). If methods like the Capabilities Approach are to be applied, there needs to be general buy-in that the need principle should be considered as a distributive justice strategy for climate adaptation and potentially prioritized over more popular principles of equality and equity to address the pressing challenges of frontline communities.

Research Question: How does the inclusion of place-based critical racial history affect support for a need-based distributive justice strategy in design for climate adaptation?

Hypothesis: People with more information about racial history will show stronger support for the need-based allocation of resources than people with less information about racial history.

This quantitative, two-cell experiment gathered data from the participants of Amazon Mechanical Turk (MTurk) living in the United States (Litman et al. 2017). As the use of MTurk for online studies has increased in popularity, so has the understanding of its benefits and concerns. For example, Aguinis et al. (2021) reviewed MTurk research across the field of management research and came up with best practices to avoid concerns affecting validity, such as participant inattention, self-misrepresentation, self-selection bias, vulnerability to web robots (bots), and social desirability bias. We followed the recommendations in this article to address as many of these concerns as possible.

In order to reduce variability in the interpretation of variables, the study context presented was urban heat vulnerability in Richmond, Virginia. RVAgreen 2050 is “Richmond’s equity-centered climate action and resilience planning initiative to reduce greenhouse gas emissions 45% by 2030, achieve net zero greenhouse gas emissions by 2050 and help the community adapt to Richmond’s climate impacts of extreme heat, precipitation, and flooding” (City of Richmond 2022). Because of these efforts, geographic information system maps of the city were already created with overlays that contain a variety of information relevant to climate equity (City of Richmond, Office of Sustainability 2021). Two overlays were used to develop the context for this study: Extreme Heat and Interstate Construction. Literature documenting the heat vulnerability challenges in the city and place-based racial history related to the construction of Interstate-95 (I-95) were used in the creation of a video vignette, which served as the independent variable (IV) (Evans 2015; Marquette 2020; Saverino et al. 2021).

The sample size range for this study was determined a priori using G*Power 3.1 software (Faul et al. 2009). The pilot data indicated that a small effect size was likely (Cohen’s d = 0.1), so this study was designed to balance the anticipated small effect with the feasibility of collecting a large sample. In order to detect an effect of 0.1 with an alpha of 0.01, 95% power, and an allocation ratio of 1, the total sample size would need to be 7,130 (3,565 per group). A sample of 7,130 was beyond the constraints of this study. Therefore, the maximum effect size that we would expect to see based on the pilot data would be the upper range of a small effect (Cohen’s d = 0.2). With an effect size of 0.2, a sample size of N = 1,786 would have enough power (1 − β < 0.95) to detect an effect (α = 0.01). Therefore, we aimed for a participant enrollment threshold of N = 2,000.

A sample of 2,197 participants was enrolled from a general pollution subject pool in the United States via Amazon Mechanical Turk. A total of 336 (228 intervention, 108 control) participants were screened out for failing to pay attention to the independent variable video in the first few minutes of the study. A total of 46 (22 intervention, 24 control) additional participants dropped out before the end of the study and 84 (43 intervention, 41 control) participants were excluded based on preregistered criteria (Gordon and Klotz 2021). The independent variable attention check filtered out approximately twice as many participants from the intervention group than the control group. A supplementary analysis that compared the demographic proportions between the intervention and the control group did not detect that this had a notable effect on the data. However, it is possible that there are unmeasured explanations for the differential attrition, so the potential contributions of this are examined in the discussion. The final sample (N = 1,731) included 900 intervention group participants and 831 control group participants. The groups were proportional on all demographics including gender identity (50.7% women), age (M = 42.7, SD = 13.4), race (78.1% white), and formal education (39.5% had a bachelor’s degree).

When designing this study, there were two types of bias that were of particular concern: desirability bias and self-selection bias. Desirability bias is a type of bias that comes about as a result of participants wanting to answer in a way that is socially appropriate. With highly political issues like race and climate change, the risk of desirability bias will always be higher than that with less-politicized topics. In order to combat this bias, the anonymity of responses was emphasized and valid reasons to support all of the strategies were presented in the video. The possibility of self-selection bias was high due to differential attrition. To limit self-selection bias, there was no indication that racial topics were involved in the study prior to the intervention. However, the recruitment materials did indicate that the study involved decision-making for climate infrastructure adaptation, which could bias the results in favor of people who are interested in that topic.

After consenting to participate, participants were randomly assigned to either the intervention or the control condition. The participants then viewed a video about a fictional climate adaptation project addressing heat vulnerability in the city of Richmond, Virginia. In the middle of the intervention group’s video, there was a 50-s clip detailing place-based racial history about the historically Black neighborhood of Jackson Ward. On the next page, all participants responded to an attention check on the video they viewed. The participants had the opportunity to go back and watch the video before submitting the question and were notified that if they did not answer the question correctly, they would be screened out of the survey. All participants who answered the attention check correctly proceeded to a question inquiring about their support for (or opposition to) a need-based distributive justice strategy. The need-based distributive justice strategy was represented visually and the participants were guided through understanding the elements on the map prior to answering the question. The participants also reported their support for (or opposition to) an equality-based strategy and a merit-based strategy, and then, they were asked to choose one of the strategies, asked a series of follow-up exploratory questions, and then asked to do a comprehension check. Finally, the participants reported their demographic information and reported any discrepancies that would make their data less usable before receiving compensation.

The dependent variable (DV) was measured on a scale from 1 (Strongly oppose) to 6 (Strongly agree). Participants were asked the question “How much do you support or oppose this distribution of resources?” (Fig. 1). The distribution of resources for the need-based distributive justice strategy was represented by monetary allocations ranging from $100,000 to $300,000 based on current heat vulnerability estimates. The full visual for this measure can be found in the Supplemental Materials.

While single-item dependent variables are quite common for the support or oppose type of question that we asked in this study, there can be drawbacks to not using a multi-item measure. Most notably, there can be a reduced sensitivity, which makes it more difficult to detect whether the hypothesized differences are truly absent or simply not detectable with a single-item measure. However, this single question best represents how this design choice may be presented in practice. Additionally, there are several other constructs, including racial preferences, which are best measured with a single item (Axt 2018).

Table 1 summarizes the collected data and analyses used. The dependent variable means were compared using Welch’s test. Because variances are rarely equal in experimental data, we elected to use the Welch’s test in lieu of Student’s t-test. This may help reduce Type 1 error (Delacre et al. 2017; Rasch et al. 2011). Additionally, Welch’s test is robust to assumptions of normality, given a large enough sample size (Rochon et al. 2012). R packages used in the data analysis include dplyr, ggplot2, tidyr, ggpubr, broom, and rstatix (Kassambara 2020, 2021; Robinson et al. 2021; Wickham 2016, 2021; Wickham et al. 2021). As a supplementary analysis, we conducted a mediation analysis using the mediation package in R to better understand the role of a perceived relevance of racial history on the relationship between the independent and the dependent variable (Tingley et al. 2014).

We hypothesized that the group exposed to racial history would show stronger support for the need-based allocation of resources than the group not exposed to racial history. The degree of support for a need-based distributive justice strategy did not differ significantly according to a Welch’s t-test, t(1,710.8) = −1.45, P = 0.07 (Table 2). The Likert scale ranged from 1 (strongly oppose) to 6 (strongly support). On average, participants in the intervention group indicated a higher mean support score (M = 4.65, SD = 1.22) than the control group (M = 4.57, SD = 1.25). The 99% confidence interval for the effect of place-based racial history on support for a need-based distributive justice strategy is between-infinity and 0.05. These results do not support the hypothesis.

A full report of the exploratory analyses can be found in the Supplemental Materials. Notable findings include the following: (1) the intervention group perceived racial history to be more relevant to their decision-making than the control group; (2) the relationship between the IV and the DV was mediated by the degree to which participants perceived the racial history presented to be relevant to the stated problem; (3) the intervention was correlated with a significant decrease in support for a merit-based strategy; and (4) support for a need-based strategy did not differ between Black and white participants.

Despite there being a shift toward perceived greater relevance at the group level, the shift occurred between completely irrelevant and somewhat irrelevant. Therefore, this suggests that, in this case, the majority of participants did not believe that the racial legacy of the highway should be tied to decision-making in the present and about the future. This result adds complexity to and possibly represents a deviation from an assumption based off suggestions from previous studies that incorporating place-focused racial history education would increase an acknowledgment of the relevance of racial history to the built environment (Bonam et al. 2018). Specifically, for engineering decisions, which have traditionally involved less consideration of more complex social factors—like place-based racial history—there may be more of a need to elucidate exactly what that relationship is and why it matters. It may also be the case that a brief intervention providing the information for a single situation is not the only intervention that would be needed to expand this perspective. For example, systems thinking techniques may need to be explored to allow individuals to make the connections for themselves before project-level information can be useful. Future work may address this dynamic by testing the interventions of varying durations and context.

A mediation analysis revealed a strong indirect effect of perceived relevance of racial history on the relationship between a knowledge of racial history and a support for a need-based distributive justice strategy and a weak, statistically nonsignificant direct effect. These results suggest that whatever effect this intervention may have on decision-making is strongly informed by the positionality of the individual and that there is potential to see stronger effects if more participants are convinced about the relevance of the racial history being discussed to the decision being made. Because history is presented through a subjective lens and can, to an extent, be accurately presented in a variety of ways, future work should interrogate which vantage points for presenting racial history are most effective for helping a diverse population connect that history to present-day built environment decision-making. This may be a challenging prospect for engineering to grapple with because of the acknowledgment that the intervention is subjective in nature. Previous work has illustrated how engineering culture can benefit from moving away from culture aligned with depoliticization and objectivity (Cech 2013). While it is important to strive for designs that accommodate all people, these aspects of traditional engineering disregard the value that comes from a more nuanced understanding of these factors. When engineers consider the social power dynamics of their work—when they consider the multiple vantage points of a historical context—the work that they do will be politically engaged in a way that can be socially responsible. Similarly, objectivity presumes that all relevant design factors that can be quantified are removed from context, which is not the case for socially responsible projects.

In addition to the data collected about need-based distributive justice, exploratory data on merit-based and equality-based distributive justice strategies were also collected. While there was no discernible effect on the equality-based strategy, the opposition to the merit-based strategy increased. The merit-based distributive justice strategy presented the exact opposite scenario of the need-based distributive justice strategy.

The merit-based distributive justice strategy was widely unpopular among both groups. However, the data showed a statistically significant increase in opposition between the groups for this strategy. Given the deep-rooted support of meritocratic ideology in US American identity, it makes sense that presenting a narrative of place-based critical racial history might further disrupt the perception that a strategy based on merit—without considering the context of the past—is unjust.

Given the strong response to the merit-based strategy in the study, future work may consider the different ways in which the concept of merit could be framed to prioritize communities in the same way that need-based distributive justice has the potential to. In this study, merit was presented based on a definition from the literature: “According to the principle of equity/merit, resources should be distributed in accordance with individual contributions” (Rothmund et al. 2016). Perhaps in future studies, it would be beneficial to test the framing of merit as a lack of contribution to the effects of climate change (i.e., a small carbon footprint), rather than as a contribution of financial resources. This collectivist framing of merit will differ from other contexts where merit can be more intuitively assessed at the individual level.

The study of denial of systemic racism in the social-psychology literature has two dominant explanations about why white US Americans perceive less systemic racism than US Americans from minority groups: an information-based explanation and a motivation-based explanation (Lesick and Zell 2020). The information-based explanation was studied in this experiment. The motivation-based explanation posits that engaging with the problem of systemic racism can result in feelings of identity threat, which activates a defensive response leading to a desire to deny the existence of systemic racism or distance oneself from it (Knowles et al. 2014). In reality, both explanations are likely to play a role. If this is the case, we would expect to see a smaller increase in support for a need-based distributive justice strategy for white intervention participants than for Black intervention participants. The data did not support this idea: the difference in means between the white intervention group and the white control group was smaller (Δ = 0.07), compared with Black participants (Δ = 0.05). However, a motivation-based influence should not be entirely ruled out because the data compared previously compares two very differently sized groups: The white group comprised 1,385 participants and the Black group comprised 160 participants.

In future studies, it may be helpful to design a more complex experiment that integrates self-affirmation (Lesick and Zell 2020), perspective-taking (Mischinski 2019), or more specific guidance about how to receive and apply information about place-based racial history to the present (Onyeador et al. 2021). While these interventions have not produced replicable effects yet, they have shown promise and are still in the process of development. After stronger interventions are developed from these ideas, they will likely be helpful in developing a more complex experiment that clarifies the effects of place-based racial history for climate adaptation decision-making.

In addition to the plans for future work outlined previously, a measure of support for climate adaptation policy that has multiple items would increase the sensitivity of the measurement and require fewer participants to detect a small effect. The most closely related scale that measures attitudes toward the universal distributive justice principles is the Basic Social Justice Orientation (BSJO) scale (Hülle et al. 2018). This scale was developed and validated in German and cannot be used with an English-speaking audience without further testing. Additionally, it is framed in terms of agreeing or disagreeing with a notion of justice, whereas this study was more interested in measuring support for an operationalized representation of the justice principles.

A subanalysis, available in the Supplemental Materials, was conducted to better understand the potential effects of differential attrition within the study. The analysis included examining the question behavior responsible for the differential attrition, the difference in video lengths, a quantitative analysis searching for significant differences between measured demographic variables, and a sensitivity analysis to assess the potential effect of unmeasured factors on the study outcome. The first three items in the analysis suggest that differential attrition probably did not have an impact on the study. In the event that an unmeasured variable did systematically filter out participants who would have responded in the same way to the dependent variable, the sensitivity analysis shows that their responses could have had a significant outcome on a statistically significant conclusion. However, since the hypothesis was not supported in this study, it is unlikely that differential attrition had an effect on the conclusions.

More empirical data are needed before concrete recommendations can be presented to practitioners. In projects that are equity-focused, it is increasingly common to include place-based racial history in the background and motivation components of project discussions. The data align with a well-known practical finding: audience matters. Within the boundaries of the previously discussed limitations, this study indicated that the place-based racial history presented in this study did not present a significant impact on the overall sample, but that it had much more of an effect on participants who found the specific content of the racial history presented to be relevant. Allowing time for dialogue and conversation specifically on issues of place-based racial history may be necessary to converge on an understanding of the implications of place-based racial history on a fair allocation of resources for climate adaptation.