Researchers Should Understand and Adapt Race and Ethnicity Data That Change Over Time
Embedding race equity principles into supports provided for young people who age out of foster care can better prepare them for a successful transition into adulthood. Child welfare practitioners and policymakers must consider how race and racism affect a young person’s child welfare experience and the services and supports they receive. For example, practitioners and policymakers should understand how employment program outcomes vary by race/ethnicity, or the ways in which access to culturally competent sexual and reproductive health care varies by race/ethnicity. This focus on race equity principles ensures that all young people have access to services tailored to their needs.
For practitioners and policymakers to accurately interpret data and make decisions about programming for all racial and ethnic groups, researchers must be able to capture someone’s racial and ethnic identity alongside their outcomes. One common resource available to researchers who want to examine outcomes over time is panel, or longitudinal, data, for which the same people are repeatedly and regularly surveyed over an extended period of time. However, researchers should carefully consider how they use these data in analysis because individuals’ responses to race/ethnicity and other demographic variables may change over time. When researchers treat race/ethnicity as an unchanging variable they potentially miss important equity considerations.
Reviews of panel data show that responses to questions on racial and ethnic identity can and do change over time. While this is a fairly common occurrence in longitudinal data for respondents of all ages (adolescence through adulthood), such changes may be particularly meaningful for young people aging out of foster care. These young people’s child welfare experiences (e.g., frequent moves, lack of information about family history, placement in foster homes with parents of a different racial and ethnic identity) may leave them without the information needed to form a healthy racial and ethnic identity. During the transition to adulthood, implicit and explicit biases around racial and ethnic identity from both individuals and systems can create opportunities and barriers at key moments in life, such as pursing postsecondary education or attaining first jobs. Despite the potential fluidity of racial and ethnic identity, however, this variable is commonly treated as static and unchanging in analysis. To date, there are few resources to guide researchers in designing and conducting analyses that both honor the racial and ethnic identities of young people and maximize the reliability of the data.
- Racial and ethnic identity formation is a normative developmental process for adolescents and young adults that typically includes many identity changes as young people explore their history.
- The child welfare system does not consistently provide young people in or leaving foster care with the historical and family information that help shape racial and ethnic identity, which may slow this developmental process for young people with child welfare experience.
- As researchers increasingly turn to longitudinal data to study the long-term outcomes that young people may experience—and with an eye toward more equitable outcomes—small decisions during data analysis (e.g., which responses to use for racial and ethnic identity when the responses vary over time) can change the final results of the analysis.
- Evidence-based policy and practice rely on study results to understand what factors have the most impact on young people’s lives and to measure progress toward equitable outcomes. When a study’s results do not show group differences, researchers, policymakers, and practitioners may conclude that those studied factors require no further attention. In contrast, a study with results showing group differences helps identify what factors to leverage in effecting change.
- Any one analysis approach is not necessarily best because the approach chosen is informed by a study’s goals. Researchers should be transparent about how their data are constructed to increase replicability and cross-study utility of results involving racial and ethnic identity.
In this brief, we first provide some background on racial and ethnic identity formation and describe some of the barriers to this identity formation process that child welfare system involvement may create for young people. Next, we qualitatively explore, through interviews with former foster youth, why racial and ethnic identity may shift during emerging adulthood, particularly among young people with foster care experience. The interviews provide context on the importance of honoring a young person’s chosen identity as that identity shifts. We then explore the practical implications of these identity changes for researchers by quantitatively demonstrating how small decisions made while preparing longitudinal data for analysis can produce completely different results.
After describing patterns of racial and ethnic changes observed in our dataset, we then undertake what we call a “three-approach analysis” in which we repeat the same analysis three different ways, with the only change being how we prepare the racial and ethnic data. We conclude by discussing the equity implications of being transparent and detailed when describing how racial and ethnic identity data is used in research studies.
The field’s current understanding of race, ethnicity, and racial and ethnic identity has evolved beyond the notion that people can be categorized into distinct racial and ethnic groups. Society’s understanding has shifted from assuming absolute genetic differences to recognizing the interplay of multiple factors in determining racial and ethnic identity. Factors that interact to shape racial and ethnic identity include phenotypical attributes (e.g., skin color, facial features), ancestral roots or common descent (e.g., where a person’s ancestors are from), geographical and social context (e.g., traditions of a region in the country), and others’ perception of racial categories and boundaries. Along with these external factors that can change over time, racial self-identification is another form of racial and ethnic identity with the potential to shift over time and across contexts.
Racial and ethnic identity formation is an ongoing developmental process
Racial and ethnic identity formation is a developmental process that begins in early childhood and sees more rapid development in late adolescence. During emerging adulthood, adolescents experience a critical period of identity formation that involves more use of abstract thinking, a more nuanced self-identity, and increased awareness of social groups. During this developmental period, young people typically grapple with identity questions and seek out multiple sources of information as they explore their ethnic and racial heritage. This process is called racial socialization, and families are the earliest and most frequent sources of racial socialization. Conversations, traditions, and exposure to community from parents and other family members shape the process of young people’s racial and ethnic identity formation by helping them learn values, activities, and behaviors associated with their racial and ethnic identity, and by preparing them for experiences with racial discrimination and stigmatization. In addition to this family context, socialization experiences and pressures—such as contact with peers from different backgrounds, exposure to discrimination, and participation in activities related to ethnicity—increase the saliency of racial and ethnic identity during adolescence and early adulthood.
Existing research has demonstrated that racial and ethnic identity fluidity varies widely among different racial groups. A study of linked, individual-level United States Census data from 162 million people from 2000—the first year in which respondents could select two or more races—to 2010 found that about 9.8 million people (6.1%) changed their race and/or Hispanic origin response. Changes in racial identity were prevalent among respondents who identified as American Indian, Alaska Native, Native Hawaiian, Other Pacific Islander, Hispanic, or two or more races On the other hand, studies suggest that racial identification for non-Hispanic White, non-Hispanic Black, and non-Hispanic Asian race groups remain largely consistent. Racial fluidity is also seen in the concept of racial “passing,” in which a person of one racial group is perceived instead as a member of another. Individuals who are mixed-race or perceived as racially ambiguous may be able to fluidly pass in this way. Stability or change in racial and ethnic identity influence how individuals perceive themselves, how others perceive and treat them, and ultimately how they navigate the world.
Child welfare system involvement can impede racial and ethnicity identity formation
As with all adolescents and emerging adults, young people with experience in the child welfare system explore their racial and ethnic identity and may experience developmentally appropriate shifts in identity. However, child welfare-specific experiences and systemic/structural issues—denied or delayed access to family history, placement in transracial or transethnic foster homes, and blocked access to relatives and other persons with shared racial and ethnic identity—may deeply complicate racial and ethnic identity formation and contribute to further shifts in racial and ethnic self-identification. Recent research has shown higher rates of fluctuation in self-reported racial and ethnic identity among foster youth than among adolescents in the general population. For example, young people with foster care experience are more likely to change their race/ethnicity as they age than their peers not in foster care. In addition to the issues discussed above, other child welfare experiences and systemic issues may contribute to disruptions in racial and ethnic identity development, including inaccurate, inconsistent, or missing labelling by professionals in child welfare; loss of family history from intergenerational foster care experience; and experiences of trauma and discrimination in congregate care and foster care settings.
Young people who have experienced transracial, transethnic, or transnational foster or adoptive placements may face challenges with racial and ethnic identity development due to shifting socialization practices and varying degrees of connection and disconnection with individuals who have shared identities. Young people who are removed from their families of origin may experience a loss of connection to familial and cultural roots that alters their racial and ethnic identity development. For example, it is critical to acknowledge the child welfare system’s legacy of systematically separating Native American youth from their home and community for placement into White homes. The Indian Adoption Act of 1958, Indian Adoption Project, Adoption Resource Exchanges system, institutionalization of Native children in remote schools, and additional policies and practices designed to dismantle American Indian and Alaska Native tribes through child removal continue to have lasting implications. The historical dimensions of the Indian child welfare crisis further highlight the loss and challenges associated with transracial and transethnic placement, as well as the importance of preserving cultural connections and providing young people with access to familial and cultural history.
In nearly all states, statutes designed to protect everyone’s right to privacy impede young people who have been adopted from accessing their own records, including birth certificates, contact information of their families of origin, medical histories (e.g., health records), and social family histories (e.g., race, ethnicity, religion). Upon turning 18, young people may request identifying information if their birth parents have consented to the release of that information. Without this consent, getting access to such information is often too difficult and cumbersome to navigate. Delayed or denied access to family information and history may leave young people without information that can teach or reinforce their racial and ethnic identity.
Longitudinal data present both opportunities and challenges in understanding identity changes over time
Many research projects are interested in understanding how people’s lives change over time and how those changes are related to some of the outcomes that people experience. These projects follow participants over an extended period (usually several years) and survey participants at regular intervals (e.g., every 6 months). This longitudinal style of data collection allows comparisons of the same person across multiple points in time, which typically offers more consistency and certainty than having to compare that person to someone else. In one prominent example, the National Longitudinal Study of Adolescent Health (also known as Add Health) started in 1994 by collecting a wide range of data from more than 20,000 young people in grades 7 through 12 during the 1994-95 school year, along with data from their school, peers, and parents. Between this first 1994 data collection and 2018, Add Health has added four more waves of data from the same young people into adulthood. This combined dataset has been used to generate thousands of published papers, and has advanced social, medical, and economic understanding of adolescent and young adult development.
While longitudinal data allow a wealth of contexts to be gathered, such data also have some complications. Researchers are interested in measuring change in certain factors, but they expect other factors to remain unchanged (e.g., biological parents or blood type)—or, at least, to change in a predictable way (e.g., age). Researchers rely on these unchanging factors to narrow down the possible reasons for why an outcome happened. For example, if two students in a study on bullying report that they live in the same neighborhood, then researchers can rule out, or control for, “living in different neighborhoods” as the reason these two students had different bullying experiences. The most common choices for these control factors are demographic variables like racial and ethnic identity, age, and sexual identity.
Problems occur when researchers expect that a person’s survey response to these control factors will not change, but the response does, in fact, change across surveys. Responses can change for many reasons. Some changes do not reflect a true change, such as an error in data entry (e.g., a participant accidentally selects an incorrect response, or a researcher mishears a participant and writes down an incorrect response) or participants—and especially adolescents—who intentionally select an incorrect response to be funny. However, the change in response may also reflect the participant’s true change. As mentioned in previous sections, racial and ethnic identity can and does change. Researchers must decide how to handle these changes when selecting which responses to use for a study.
Young People with Foster Care Experience Highlight Causes of Racial and Ethnic Identity Shifts
We conducted interviews with six young people who have participated in the Jim Casey Youth Opportunities Initiative to learn more about racial and ethnic identity development. The Jim Casey Initiative aims to ensure that young people ages 14 to 26 who have spent at least one day in foster care after their 14th birthday have the resources, relationships, and opportunities to thrive in adulthood. The Jim Casey Initiative currently operates in 16 states across the country and runs Opportunity Passport, an asset development and financial literacy program. Young people involved in the Jim Casey Initiative complete a survey when they enroll in Opportunity Passport, and again every April and October while they are active in the program. Our six young interviewees were all formerly or currently enrolled in Opportunity Passport. Of these young people, four changed their responses to survey questions about racial and ethnic identity over time, and two did not; all were over age 18 and identified as a person of color. We utilized a semi-structured interview guide to conduct a one-hour virtual interview via Microsoft Teams with each young person. After all interviews were complete, we uploaded and coded the transcripts in Dedoose using a coding guide with themes identified a priori (i.e., before coding began), based on the relevant literature available. These themes were updated as the research team coded the data. The following sections present the themes that emerged from three topics: participant characteristics, reasons for changing racial and ethnic identity, and the meaning and permanency of racial and ethnic identity changes.
Interview Topic 1: Participant characteristics
Current and previous racial and ethnic identity
The most common change throughout the interviews was for young people to change from indicating one racial and ethnic identity to indicating more than one racial and ethnic identity. This change was most often the result of young people learning more about their family of origin or ancestry. Young people also reported that learning more about what racial and ethnic identity meant to society and to themselves led them to report a different identity across different rounds of the survey. Young people reported that their own racial and ethnic identity often shifted after they learned about race and ethnicity as constructs rather than discovering new pieces of their own identity.
Influences on racial and ethnic identity formation process
Young people reported a number of sources that influence their current and previous racial and ethnic identity. These sources included school, child welfare, family and/or friends, and media, including social media. Several young people reported that high school and post-secondary education experiences shaped their racial and ethnic identity. Young people spoke about teachers’ positive and negative influences on their identity. Their experiences ranged from professors at Historically Black Colleges and Universities (HBCU) helping young Black people feel connected to their culture and proud of their identity to high school teachers making racist comments in the classroom that caused young people of color to question their own identities and cultures. Young people also spoke about the racial composition of their school and the importance, to their own identity development, of having peers at school who looked like and accepted them.
- “This professor helped in making me feel as though it was time to claim who I am as a whole person. You know, [you] don’t claim one part, [you] don’t claim half of it, you know, [you] don’t only claim it when it’s, you know, necessary for a scholarship or necessary for an opportunity. [I] claim it every day of my life and wake up in on that.”
- “[After moving several states to be with a family member], when I went to that public school in [my state], it was like a game changer for me. I felt like I was learning how to walk all over again. […] Like I was forced to learn about my culture because that’s what was all around. I had to learn how to speak Spanish.”
The child welfare system played a large role in shaping the identities of several young people we interviewed. Young people discussed the ways in which the child welfare system disconnected them from their culture and exposed them to people of different races/ethnicities but did not teach them about their own identity.
- “I was adopted around nine [years old], and I moved in with a very different lady. ‘Cause I grew up in a, like, Mexican family [and] culture, and then when I got adopted, there was a huge shift. I stopped going to bilingual class. I stopped all the traditions that we were doing, and that really changed my whole [identity to the point that I thought], well, maybe I’m not Mexican. I’m just plain White ‘cause of those experiences [after adoption].”
- “When everything happened with foster care, I kind of became ashamed because of my culture. […] You become very ashamed of your identity and your culture because you’re in a system where it’s always, ‘You guys did something wrong. It’s your culture: you guys are lazy.’”
- “[My caseworker] definitely was a huge inspiration when it came to just remembering that my Blackness is still beautiful. You know, even though I was going through the system, even though I had those bad days, those dirty days, those down days, you know, I was still beautiful inside, and that’s what reflected outside.”
Young people reported that family was influential in their racial and ethnic identity development and provided an understanding of their past and ancestry. The young people described both positive influences—in which family members were positive role models and taught young people about their culture—and negative influences, in which families did not want to accept certain aspects (e.g., only acknowledging one of a multiracial young person’s races/ethnicities) of the young person’s racial and ethnic identity.
- “[My aunts] were the most beautiful people to me. One of my aunts experienced a lot of racism, even though she was lighter skinned. She knew who she was and owned that. I looked up to them when I was growing up.”
- “You know, when I wasn’t accepted by my father’s family, it was hard to identify as who I am because it’s like the part of me that is the most Black, you know, doesn’t even accept me.”
While four of the six young people reported that friends had influenced their racial and ethnic identity, this influence was not as strong as the influences from family or the other sources mentioned in this section (school, child welfare, etc.).
- “In middle school, my best friend was Asian American. When we got older, her mother and my mother did not get along, and [her mother] made race-related comments against us. That made me question: Is my Black good enough? Am I going to be able to make friends outside my race?”
Young people reported that media was another important influence on their racial and ethnic identity development. As with the other influences, young people reported both positive and negative influences. The respondents mentioned several types of media as influencing their identity, including social media, movies, and television. Some young people reported that media had positively impacted their identity development by expanding their worldview and providing them with the racial and ethnic representation not found in their communities. Others reported that media was a negative influence because of negative stereotypes portrayed.
- “[Television] helps you see. I hate to make it all sappy, but you see what you’re capable of if you see, you know, a positive racial influence on television.”
- “I saw this movie called Towelhead. It was about this Arab American girl living in the suburbs in California. They would call her all kinds of names. […] Moving countries like that [girl in the movie], I thought maybe that’s how my dad felt. And then I thought about how the girl was treated and didn’t want to be treated like that.”
Interview Topic 2: Reasons for changing identity
In addition to seeking to understand the sources of influence on a young person’s racial and ethnic identity development, we asked the young interview respondents to share their perspectives on why they or other young people might report changes in their identity over time. Young people attributed these changes to learning more about their racial/ethnic ancestry, their change in self-view and others’ views of them, the child welfare system, and societal pressures. One main reason for why young people changed their racial and ethnic identity was learning new information about their ancestry. Sometimes this information came from family members; other times, it was discovered through a DNA test. Young people reported:
- “You know, taking ancestry tests to just confirm or deny [ancestral racial and ethnic identity], you know, and seeing what the truth really was. Those helped [form my race/ethnic identity] as well.”
- “[If] you discovered something new, and you want to claim that heritage, then that’s more power to you because that’s who you are.”
In addition to learning new information about their ancestry, young people also reported changing their racial and ethnic identity due to changes in how they viewed themselves or in how others viewed them. Changes in young people’s overall self-perception affected their racial and ethnic identity formation, which included learning more about themselves, gaining confidence, and addressing internalized racism. Young people reported:
- “Sometimes your own race makes you hate certain things, makes you think that, oh yeah, it’s not beautiful, or it’s not okay to be a certain way. So, you end up hating yourself, or you end up just not seeing the beauty in [your race].”
- “I put myself on a pedestal. So I say, you know, where my Blackness is above all, you know, everything that flows through this universe comes from my beauty of Blackness.”
Three-Approach Analysis Reveals the Impact of Small Coding Changes on Analysis Outcomes
To demonstrate how survey development and data analysis decisions can influence results when including data that could change over time—such as racial and ethnic identity—we conducted the same analysis three different ways. In using these three approaches, we mirrored common decisions made when collecting and preparing panel data for analysis: 1) reporting on race/ethnicity only during the first round of data collection and assuming it does not change, 2) assuming that the most frequently reported response is most reflective of the individual’s race/ethnicity, or 3) assuming that the most recently reported response is the most reflective.
Jim Casey Youth Opportunities Initiative restricted use dataset
The Jim Casey Initiative administers the Opportunity Passport Participant Survey (OPPS) to young people at program entry and then twice a year (April and October) across all Jim Casey Initiative sites. Young people are first asked whether they identify as Hispanic or Latino and then asked to check all options that apply for their racial background, from the following: White, Black/African American, Asian, Native Hawaiian/Part Native Hawaiian, Other Pacific Islander, Native American/Alaska Native, or Other.
For these analyses, we looked at individuals who classified their racial and ethnic identity as any options other than “Other” on at least two surveys. To protect the privacy of young people and prevent reidentification of individuals, some racial and ethnic identity information was masked. Masked racial and ethnic identity information was combined with the “Other” category, which is self-selected by young people. For example, due to the small number of surveys in which a young person identified as Native American or Alaska Native, non-Hispanic, all of these instances were recategorized into the Masked/Other category.
Participant-level change patterns
Of the 1,967 participants in the restricted use dataset who have completed at least two surveys, 329 (or 17%) changed their racial and ethnic identity at some point. We categorized these participants into four change patterns: sustained, reverted, alternating, and fluid. The categories are defined below in the figure. Sustained identity changes were the most common, followed by reverted changes, alternating changes, and fluid changes. The majority of the change patterns included young people who identified as “Two or more races, non-Hispanic (NH)” at one point in time.
Survey-level change patterns
In the previous section, we examined participant-level change patterns, which tell us the overall change pattern for a participant across their surveys but does not tell us which racial and ethnic identities were involved, or how many changes occurred for alternating and fluid changes. To find these answers, we also looked at survey-level change patterns, which compare one survey to the next. There were 703 surveys taken across 329 participants for which the subsequent survey had a different racial and ethnic identity, meaning that many participants had multiple racial and ethnic identity changes. The most common identity change was from a single identity of color to another identity of color—the majority of which involved a change to or from “Two or more races, NH.” When very few participants report a racial and ethnic identity category in the dataset, we mask that data (i.e., we do not report those frequencies) to protect their privacy; combining these categories in one “Masked/Other” category allows us to ensure that all young people of color are captured in our analyses.
We constructed three versions of the dataset by replacing all of a participant’s racial and ethnic identity responses with the identity reported on 1) their first survey, 2) their most commonly reported racial and ethnic identity, and 3) their most recent survey.
- The first response version of the dataset applies the racial and ethnic identity a participant reported on the first survey they took to all of their surveys. This approach simulates datasets for which racial and ethnic identity is only asked the first time the survey is taken.
- We also created a version of the dataset containing the racial and ethnic identity that was the most common response across all surveys taken. In instances where there were multiple equally most common responses, we broke ties using an ordered list that prioritizes the least commonly reported racial and ethnic identities to capture as many instances of those identities as possible. This approach assumes that the most representative identity is the race/ethnicity the participant most often reports.
- The most recent response version of the dataset contains the racial and ethnic identity that a participant reported on their most recently taken survey. This assumes that a participant’s most recent response is more representative of their identity than prior responses.
Among participants who changed their reported racial and ethnic identity at some point, the racial and ethnic identity breakdown changed in the following ways depending on the dataset construction method.
Number of foster care placements by racial and ethnic identity
To demonstrate how each of the three approaches may affect child welfare characteristics within a sample, the table below demonstrates the differences in number of placements by racial and ethnic identity. There is not a large difference in counts depending on the approach: Only 17 percent of survey participants changed their race/ethnicity response across surveys.
* F = First Response; MC = Most Common Response; MR = Most Recent Response
One-way analysis of variance (ANOVA)
The one-way analysis of variance (ANOVA) examines statistically significant differences in the means of three or more unrelated groups—in this case, groups with different racial and ethnic identities. This analysis does not tell us which groups are different from each other, but just that at least two groups are different. We utilized ANOVA to examine how the results of a statistical approach change based on how racial and ethnic identity is categorized. As we show below, the choice of approach affects the results, even though all three approaches were similar in the relative size of groups.
- First response: We examined the relationship between number of placements and racial and ethnic identity when using the first racial and ethnic identity a young person reported. This did not yield a statistically significant difference between groups (F=1.77; p>.05).
- Most common response: We examined the relationship between number of placements and racial and ethnic identity when using the most common racial and ethnic identity a young person reported. This did not yield a statistically significant difference between groups (F=1.9, p>.05).
- Most recent response: We examined the relationship between number of placements and racial and ethnic identity when using the last racial and ethnic identity a young person reported. This did yield a statistically significant difference between groups (F=2.43, p<.05).
Note: *p <. 05
During our interviews, young people reinforced the notion that racial and ethnic identity development is an ongoing process. We interviewed four young people who changed their racial and ethnic identity and two young people who did not change their identity. All interviewees had foster care experience and were age 18 or older. Young people in both groups identified several influential sources in their own racial and ethnic identity development—and for race/ethnic identity development in general—including school, family and friends, media, and (notably) the child welfare system.
“[Your racial and ethnic identity] affects the way you think, the way you operate, the way you view things, the way you view life actually. […] Ethnicity makes it more challenging because people just look at you, and they just stereotype you.”
Child welfare involvement influenced race/ethnic identity development, both positively and negatively. A central theme to these young people’s experiences was the repeated culture shock that can happen with multiple placements. Some of these experiences were affirming of young people’s identities and cultural ties, while other experiences isolated and stripped them of the opportunity to form a concrete cultural connection to a chosen racial and ethnic identity. One young person talked about the isolation that comes with growing up in foster care and how it “affects the way you think, the way you operate, the way you view things, the way you view life,” with the added layer of navigating ethnic discrimination and stigmatization.
“I lived with many different families that had many different cultures and identities and race. And I just never knew where I fit in anymore. I still have a hard time with it, ’cause I still don’t know what is me, and it’s kind of hard.”
Young people who changed their racial and ethnic identity described some of their reasons for making the change. These young people stated that their changing racial and ethnic identity came from a shift in how they viewed themselves and the world, showing the effects of their formative experiences. They spoke to the importance of a strong, positive racial and ethnic identity. Given the multiple sources of influence for young people, experiencing longer stays in foster care and disruptions through multiple or culturally incongruent placements can destabilize young people’s ability to achieve the foundation of a positive racial and ethnic identity.
The three-approach analysis demonstrates how different analytic methods can change the results—and conclusions—of a study. This analysis used ANOVA to determine whether there were racial and ethnic group differences in the number of foster care placements that young people experienced to provide an example using a commonly tracked outcome of the child welfare field. Studies of the number of foster care placements can measure and track progress toward more equitable placement experiences for all young people. In the dataset we used, almost 17 percent of survey participants changed their race/ethnicity response across surveys. It would not be unreasonable to dismiss this number of changes and assume that results would be unaffected by an event that fewer than one fifth of the study sample had experienced. Indeed, when comparing counts of placement changes for each racial and ethnic group, there is not much difference among the three approaches at first glance. We can see, however, that the approach chosen does result in statistically significant differences, which is the metric researchers use to decide whether group differences have more than just a random chance of being real. The small shifts in racial and ethnic group sizes produced by each approach can be enough to detect differences.
Using a participant’s most recent response (Approach 3) may yield different results from the other two approaches for a variety of reasons. As young people figure out what identity most closely matches how they feel, a young person’s most recent response may reflect comfortable affirmation of an identity. Alternately, since other people’s perceptions play a role in both how a young person may identify and in how that young person is treated by society at large, this approach may somehow capture more of the connection among racial self-classification (which response someone checks on a survey), observed race (what race other people would classify someone as), and reflected race (what race someone thinks other people would classify them as). Qualitatively following up on these response changes may shed more light on the meaning behind them.
We do not make a recommendation on which approach is best because that decision depends on a researcher’s goals and study design factors. For example, a study that examines the effects of early life decisions and perspectives on future outcomes may find it most useful to use a respondent’s first response. Carrying the first response forward is the most traditional approach. In contrast, if a researcher is most interested in understanding how more recent experiences have shaped a person’s outcomes, it may be best to use the most recent response. If a researcher has reason to believe that changed responses are relatively rare and likely to be data entry errors instead of meaningful data, selecting the most common response can reduce some of the concerns about data quality. Sometimes, the approach has been selected before a researcher even gets the data. Secondary data analysis—using existing data collected by someone else, such as federal data—with longitudinal survey data that only asks about racial and ethnic identity at the first round of data collection is essentially deciding to use the first response. For these reasons and many others, researchers should continue their focus on transparently making and explaining the best study design decisions for their purposes.
To better control what changes are affecting the outcome being studied, researchers typically carry a participant’s first response forward to serve as a baseline measurement for analysis. In our three-approach analysis, using either a participant’s first (Approach 1) or most common response (Approach 2) gives results that are not statistically significant. Researchers using either of these approaches would conclude that there are no meaningful differences in number of foster care placements among racial and ethnic groups. In contrast, using the response that a participant gave on their most recent survey (Approach 3)—likely the closest reflection of the participant’s current racial and ethnic identity—results in significant differences across racial and ethnic groups. These researchers would likely further explore how the groups differ and what factors can effect change for each group. Researchers may then use those results to shape policy or practice by passing results on to decision makers to advance racial and ethnic equity in foster care outcomes. Beyond yielding inconsistent results for a single study, these discrepancies could make it difficult for other researchers to reproduce the findings of past studies. What is clear is that how researchers choose to handle race/ethnicity data matters.
Data is an integral tool in addressing racism and advancing racial and ethnic equity within the child welfare system. Among other elements, we must understand how people’s chosen racial and ethnic identities are being represented in the data. Researchers make numerous decisions when preparing data for analysis, including how to handle racial and ethnic identity data. The consequences of these often-opaque decisions are magnified in panel studies because researchers must decide how to handle inconsistencies in participant responses over time.
For studies that involve longitudinal surveys of young adult respondents, inconsistency in self-reported racial and ethnic identity may simply reflect the typical developmental process. Young people solidify their racial and ethnic identity throughout their youth as they explore what race means to them, to other people in their lives, and to society. For young people with foster care experience, however, aspects of their racial and ethnic identity development typically discovered through years of interactions with family members may be unknown until much later in adolescence or young adulthood. Shifts in self-reported racial and ethnic identity—particularly to and from an identity of color—tell an important story for these young people.
In the absence of clarity on how researchers handle changing responses to race/ethnicity questions within a study, we may be making faulty comparisons across studies that paint an incomplete picture of participants’ experiences. Without a clear understanding of how race and ethnicity are categorized within a study, we cannot honor young people’s identities. Furthermore, researchers and those who use research findings (e.g., policymakers) might miss important strengths or disparities across racial and ethnic groups.
This research was funded by the Annie E. Casey Foundation. We thank them for their support but acknowledge that the findings and conclusions presented in this report are those of the authors alone and do not necessarily reflect the opinions of the Foundation.
We thank the young people with foster care experience who shared their experiences in interviews and reviewed the qualitative sections crafted from those interviews. Additionally, we thank several colleagues who provided input on early drafts, including Karin Malm and Porsche Boddicker-Young. We also thank our colleagues at the Annie E. Casey Foundation who provided input on the project and a review of the manuscript, including Sandy Wilkie and Jeff Poirier.
Flannigan, A., Rosenberg, R., Liehr, A., Dalela, R., & Sanders, M. (2022). Researchers should understand and adapt race and ethnicity data that change over time. Child Trends: https://www.childtrends.org/publications/researchers-should-understand-and-adapt-race-and-ethnicity-data-that-change-over-time
 The Census combines “American Indian“ with “Alaska Native“ into one racial group and “Native Hawaiian“ with “Other Pacific Islander“ into one racial group. Here, we list the identities separately as the cited authors did to acknowledge that, while the grouped identities can overlap for some people, they are not interchangeable identities and do not represent a single monolithic identity.
 The priority list used to break ties between the most common response is as follows: Native American or Alaska Native NH, Hispanic, Native Hawaiian or Other Pacific Islander NH, Asian NH, Black NH, Two or more races NH, and White NH. For example, if four surveys were taken—two that indicated Native American or Alaska Native NH and two that indicated White NH—the racial and ethnic identity that was used for this method was Native American or Alaska Native NH.
 Data for Asian, NH and American Indian and Alaska Native, NH young people are censored due to small sample sizes.
 Data that are censored due to small sample sizes are represented by a dash.