By Krystal Cohen

Abstract

Gifted and talented programs in the United States have been an object of controversy for decades, with many arguing that gifted education widens the gap between high achieving students and their peers, typically along racial lines. There is currently a large body of literature on underrepresentation in gifted programs for Black and Latinx students, as well as low-income students, however academic research on the impact of such programs, especially for disadvantaged populations, is a far less developed research space. Drawing on data from the National Longitudinal Survey of Youth of 1979, this study utilizes propensity matching and OLS regression to examine racial and socioeconomic disparities in the long-term outcomes of participation in gifted programs. I find that: race and maternal education are significant predictors for gifted program participation, and gifted education is positively associated with achievement test scores, academic attitudes, and self-perception, with greater academic differences for non-Black/Hispanic students and students of higher socioeconomic status, and greater social-emotional differences for Black/Hispanic students and students of lower socioeconomic status.

---

Introduction

Gifted and talented programs were originally designed to ensure that high-performing students reach their full potential and are not restricted by the bounds of a traditional classroom. However, these programs, as well as ability grouping and academic tracking more broadly, have been an object of debate in the United States for decades. Some believe that accelerated programs allow academically gifted students to flourish in a challenging environment, while others question the equity of such programs and the extent to which they are effective. There is currently a large body of literature on underrepresentation in gifted programs for Black and Latinx students and the mechanisms by which academic tracking and potentially biased processes for gifted identification negatively affect students of color and low-income students (McBee 2010, 291-292; McClain and Pfeiffer 2012, 77-78; Wagner, Dymes, and Wiggan 2017, 877-878). 

Disadvantaged populations are held to standards that favor White students in the gifted identification process, contributing to higher rates of misclassification for students of color (McBee 2010, 261-263). With such large barriers to entry into gifted and talented programs for minority students, it poses the question — to what extent do gifted programs boost student outcomes, or in other words, to what extent does the exclusion of students of color contribute to the racial achievement gap? Moreover, do the racial disparities apparent in the gifted identification process carry over into student outcomes?

Academic research on the long-term impact of gifted programs, especially for disadvantaged populations, appears to be an underdeveloped research space. What happens to the students who are granted the opportunity to enroll in gifted programs despite barriers to entry? Do they fare better than their similarly capable counterparts who were not placed in gifted programs? Do they fare as well as, better, or worse than more privileged students in the same programs? This study aims to build on existing literature by exploring the relationship between a student’s demographic background and the short and long term outcomes of participation in gifted and talented programs. Drawing on data from a national longitudinal study of students in the United States, this study will explore differences between students of similar demographics and academic abilities who participated in gifted programs and those who did not (ex. Black students in regular classes vs. Black students in gifted programs), as well as differences between students of different demographics who did participate in gifted programs (ex. Black students in gifted programs vs White students in gifted programs). More specifically, this study seeks to answer the following three research questions:

• To what extent do demographic variables affect the probability of participation in gifted and talented programs?
• Does participation in a gifted and talented program improve students’ academic and social-emotional outcomes in the long term?
• Do the outcomes of gifted students differ by race/ethnicity and socioeconomic status?

These findings have significant implications for equity in public education and beyond, offering insights on K-12 education policy and gifted program design. Based on my results, policy recommendations include: a redesign of the identification process to increase representation in gifted programs, enhanced teaching training centered around fostering inclusive learning environments, and continued assessment of the experiences of students in gifted programs, especially those from underrepresented backgrounds.

Literature Review

Academic literature discusses giftedness in two primary ways: as a characteristic of a student (i.e. gifted children) and as a pedagogical approach (i.e. gifted programs). While there is overlap between the population of students who meet the criteria for giftedness and those who participate in gifted and talented programs, it is critical to the purposes of this study to distinguish between the two groups. Gifted students are defined as individuals who demonstrate extraordinary intellectual ability and key behavioral traits like creativity, curiosity, and emotional intensity (McClain and Pfeiffer 2012, 78; Guzy 2018, 13). In contrast, gifted and talented programs housed in elementary schools or other educational institutions are typically offered to a select group of students who meet special eligibility requirements. Gifted programs [1] typically consist of two fundamental components: (1) continued exposure to higher-achieving peers and (2) a more advanced curriculum that allows for deeper exploration of academic content (Bate and Clark 2013, 50; Bui, Craig, and Imberman 2014, 32).

Due to this systematic exclusion of minority students, gifted programs may exacerbate the racial achievement gap by further boosting outcomes for more privileged students while their minority peers continue to lag behind.

Gifted Identification & Underrepresentation

The most prominent approaches for identifying gifted students have historically led to the underrepresentation of disadvantaged populations, specifically students of color, and students from low-income backgrounds. As of 2012, sixteen states had no standardized decision-making policy for gifted identification and of those who did, the majority mandated that schools use intelligence tests (16 states), achievement tests (17 states), and/or teacher and parent nominations (13 states) to identify gifted students (McClain and Pfeiffer 2012, 68-69). Through a multilevel path analysis of gifted identification patterns, McBee (2010, 291-292) found that Black and Hispanic students, as well as students who qualified for free or reduced lunch, were less likely to be identified. One explanation of why underprivileged students may be at a disadvantage is teacher bias in the nomination stage of the identification process. Studies have shown that a nomination requirement for students to move into the testing stage result in a large proportion of gifted students being missed (McBee, Peters, and Miller 2016, 261-263), and due to teacher bias, many of the students being passed over are from minority groups (Ford, Grantham, and Whiting 2008b, 291-294). McBee (2010, 284) argues that although schools may equip teachers with standardized definitions of giftedness, a student’s cultural background and socioeconomic status can heavily influence how they express gifted characteristics and behaviors, and teachers who do not share the same social identities as their students may have trouble recognizing them as gifted. In addition, minority students and low-income students of similar academic and cognitive ability as their peers continue to underperform on racially biased standardized exams, such as those administered during the evaluation process for gifted programs (White et al. 2016, 11-12; McBee 2010, 284; Grodsky and Warren 2008, 392-393). Due to this systematic exclusion of minority students, gifted programs may exacerbate the racial achievement gap by further boosting outcomes for more privileged students while their minority peers continue to lag behind.

Academic and Socioemotional Outcomes of Gifted Students

Existing research exploring the academic outcomes of gifted students have produced mixed results, with many arguing that gifted and talented programs — and ability grouping models more generally — have a positive effect on student achievement (Kulik and Kulik 1982, 425; Delcourt, Cornell, and Goldberg 2007, 371-372; Smith, LeRose, and Clasen 1991, 83) and others contending that such programs have no significant impact on performance (Bui, Craig, and Imberman 2014, 52). By conducting a two-year longitudinal study on gifted elementary school students across ten different states, Delcourt, Cornell, and Goldberg (2007, 371-372) found that students placed in specialized programs for gifted students had substantially higher standardized test scores than their gifted peers who did not participate in these programs. In contrast, through a regression discontinuity analysis, Bui, Craig, and Imberman (2014, 52) found that marginal students just above the cutoff for gifted services neither improved nor worsened academically due to participation in the gifted program, however it should be noted that students just below the cutoff had access to alternative enrichment programs. In addition, several questionnaire-based studies have found that students in grouped classes exhibit more positive attitudes toward schooling and increased academic expectations of themselves and others (Bate and Clark 2013, 52). Others have found that isolated interaction with higher-achieving peers is associated with decreases in academic self-concept and perceptions of ability, especially in the early stages of participation, due to negative reference group effects (Preckel, Gotz and Frenzel 2010, 452-453; Delcourt, Cornell, and Goldberg 2007, 377). Together, these findings suggest that although generally positive, the impact of gifted and talented programs may vary based on school and student characteristics, which the present study hopes to further explore.

Like the research on academic outcomes, research on the social and emotional outcomes of students who participate in gifted and talented programs is limited and lacks general consensus. Some academic literature suggests that gifted programs, especially those following an isolation or withdrawal model, can act as a haven and help boost socioemotional outcomes for gifted students (Bate and Clark 2013, 53; Gross 2016, 420-421), and contradictory literature suggests that gifted programs may hinder the development of socioemotional skills, leading to issues with socialization (Tomlinson-Keasey and Smith-Winberry 1983, 37-39; Vialle, Heaven, and Ciarrochi 2007, 577-578). There are also several studies that have found no significant difference between the happiness and ease of socialization for gifted children and their non-gifted peers (Zeidner 2020, 9-11). Since each study focused on slightly different metrics of social-emotional well-being (i.e. happiness, self-esteem/confidence, experience socializing with others) and was based in different countries and cultural contexts, it’s not surprising that results were mixed. Another possible explanation for the discrepancy is the difference in time frame, with some studies focused on the immediate reflections of students while participating in gifted programs and others focused on the effects in the long term. This study seeks to contribute to previous research by analyzing metrics in the long term for gifted students in the United States context, specifically, with the goal of offering state-level recommendations for improving outcomes. 

Experiences of Gifted Students from Disadvantaged Backgrounds

While the majority of studies on gifted and talented programs seek to capture the outcomes of students generally, several studies have focused their attention on the experiences of gifted minority students, offering insight into how gifted programs may impact these students differently than their more privileged peers. Through a qualitative study of gifted Black students in two Ohio school districts, Ford, Grantham, and Whiting (2008, 222-223) found that these students faced negative peer pressures and were often teased for performing well in school, contributing to attitude-behavior discrepancies and underachievement. Gifted minority students may also face stereotype threat and feelings of “fewness”, or the belief that they are one of the only students of their cultural group participating in advanced programs (Crawford, Snyder and, Adelson 2020, 61-62). These experiences directly contradict literature on gifted students generally that contend that students who are labeled gifted are treated more favorably than those who are not and have a more positive sense of self. Similarly, education researchers have found that the experiences of gifted students from racial minorities may be different from the experiences of their peers due to a weaker support network of family, teachers, and school staff. Due to lack of preparation both in educating gifted students and on the topic of multicultural education, gifted program teachers may have difficulty understanding and relating to culturally diverse students, contributing to lower teacher expectations of these students and weaker student-teacher relationships (Ford, Grantham, and Whiting 2008b, 291-294; Crawford, Snyder and, Adelson 2020, 52-54). Moreover, Anderson (2020, 92) found that White teachers in gifted programs were less likely to praise gifted Black girls for their work, and even worse, would sometimes ridicule and be skeptical of them. Overall, the literature suggests that the mechanisms that contribute to positive outcomes for gifted students generally (i.e. labeling effects, higher teacher expectations, and a higher quality educational experience) may operate differently for students from disadvantaged backgrounds, in some cases, negatively influencing minority and low-income gifted students. Through a quantitative analysis of national survey data, this paper will first analyze the impact of gifted and talented programs, with a special focus on underrepresented student populations. Next, the paper will offer policy recommendations for (1) increasing representation in gifted and talented programs and (2) strengthening the support network for gifted minority and low-income students in order to boost outcomes.

Data and Methods

This study utilizes data from the National Longitudinal Survey of Youth of 1979 (NLSY79), more specifically the Child and Young Adult cohort (NLSCYA). NLSY79 originally included a nationally representative sample of 12,686 men and women who were all between the ages of 14 and 21 years old. The NLSY79 Child and Young Adult cohort follows the biological children (N = 11,530) of the women in this original study, including biennial interviews with the children and families conducted from 1986 to 2016 (U.S. Bureau of Labor Statistics 2020). This database allows for several unique benefits. Because Black and Hispanic respondents were originally oversampled, there is sufficient data for subsample analysis and comparison of gifted identification and long-term outcomes across racial and ethnic groups. However, the use of propensity matching allows for claims to be generalized to the broader United States population despite oversampling.

This study specifically focuses on a sample of children and young adults born between 1990 and 2000, who were attending elementary and/or secondary school between the years of 2000 and 2016. Among the 3,030 students who fit this age criterion, 896 were identified as gifted at some point in this time span. For the purposes of this study, gifted students were defined as those who participated in a gifted and talented program or had a special class for advanced work in the past year. See Appendix for a full demographic summary of the sample. 

Factors of interest include demographics: race/ethnicity, gender, highest grade level completed by the mother, and birth year. Maternal educational attainment was analyzed as a numeric variable (i.e. number of years of schooling) and converted to a categorical variable with the following levels: attended college, received a high school diploma, and did not complete high school. Dependent variables of interest fall under two main categories: long term academic outcomes and long term socioemotional outcomes. Since most children are identified as gifted and participate in gifted and talented programs in elementary school (Guzy 2018, 12), this study seeks to measure the outcomes of gifted and nongifted students from the point at which they transitioned to high school onward, with special attention paid to the teenage years directly following elementary school.

Analytic Strategy

This study employs several quantitative analysis approaches to exploring the above research questions. To examine the association between demographic variables and participation in a gifted and talented program, I utilized the following logistic regression model for gifted participation:

(1)       Logit (gifted_i | I_i, E_i, X) = β₀ + β₁I_i + β₂E_i + β₃I_iE_i + Xβ + ε₁

where i represents an individual student, gifted_i represents a binary metric of participation in a gifted program (1 being yes, 0 being no), I_i represents student race/ethnicity, and E_i represents maternal educational attainment (a proxy for socioeconomic status). In acknowledgement that many demographic predictors can act as confounding variables in the context of schooling, I_iE_i represents the interaction between race and maternal educational attainment. X is a vector of additional student characteristics that may influence gifted program participation, including gender and cognitive ability test scores at age 5. Finally, ε₁ represents the error term for the model. 

Since students participate in gifted programs for varying lengths of time, which may influence treatment effects, I also ran a multiple regression model for duration of participation in a gifted program:

(2)       Y_i = β₀ + β₁I_i + β₂E_i + β₃I_iE_i + Xβ + ε₁

where Y_i represents gifted participation duration (in years) for those who were ever exposed, and the variables described above in Equation 1 remain consistent. It should be noted that due to missingness across years and the biennial frequency of surveys, this variable represents a minimum number of years of participation based on available data, rather than a precise measure.

To analyze the academic and social-emotional outcomes of students in gifted and talented programs, I isolated the treatment effect of gifted program participation by employing a nearest neighbor propensity matching approach. Because students who participate in gifted programs on average are different from students who do not, I utilized propensity score matching to obtain more credible causal estimates of gifted education. 592 pairs of students were successfully matched by birth year, race/ethnicity, maternal educational attainment, and measures of cognitive ability at age 5. For both unmatched data and matched data, I utilized the following linear regression model for each continuous outcome variable:

(3)       Y_i = β₀ + β₁Z_i + β₂I_i + β₃E_i + Xβ + ε₁

where Y_i represents the continuous outcome variable, Z_irepresents gifted education, I_i represents student race/ethnicity, E_i represents maternal educational attainment, and X represents a vector of relevant student characteristics including gender and equivalent measures at age 5, if applicable. Due to the wide variability in the NLSCYA dataset, this approach seemed like a reasonable and feasible way to reduce bias due to confounding variables. Demographic factors like race and socioeconomic status may influence both student outcomes and whether they are identified as gifted. By matching gifted [2] students to nongifted students of similar cognitive abilities and backgrounds, this form of analysis allowed for a direct comparison of outcomes between groups (Dehejia and Wahba 2002, 151-161; Stuart 2010, 4-8). Lastly, I conducted subset analyses on outcomes by race/ethnicity and maternal educational attainment to examine heterogeneity in treatment effects.

Findings

Gifted Participation by Student Background

To examine the extent to which demographic variables affected identification for gifted and talented programs, I ran a logistic regression model estimating the effects of various predictors, showing that the probability of Hispanic students participating in gifted programs is only 0.7 times that of non-Black/Hispanic students. In Model 2, maternal educational attainment was added as a predictor, revealing that students whose mothers attended college were 2.4 times as likely to participate in a gifted program than those without a high school diploma. In addition, the association between race/ethnicity and gifted identification was no longer statistically significant after controlling for maternal education, suggesting that observed racial gaps may have been primarily driven by confounding socioeconomic gaps. Models 3 and 4 include interaction variables between race/ethnicity and maternal education, as well as several other predictors: gender, PIAT math score at age 5, PIAT reading recognition score at age 5, PIAT reading comprehension score at age 5, and a dummy variable indicating missing test scores. [3] Both models showed that the probability of male students being identified as gifted is only about 0.8 times the likelihood for female students. In addition, math test scores and reading comprehension test scores were significant predictors, both of which are commonly used gifted identification criteria. 

Surprisingly, being Black did not appear to have a statistically significant effect on gifted identification, as found in prior literature on gifted identification. This suggests that Black-White racial disparities in gifted education may be more prominent at the school or district level or when utilizing cross-sectional data (on which most previous studies focus) than at the aggregated national level. Another potential explanation is that the disparities are driven by differences in gifted program persistence and how long students remain in programs as suggested by McBee (2010, 284) and as supported by findings on predictors of gifted duration.

Because the length of participation in a gifted program may impact outcomes, I also estimated the effects of various predictors on the duration of gifted program participation for those ever identified as gifted, using the same four models (see Table 2). Model 1 showed that Black and Hispanic students identified as gifted participate in gifted and talented programs for a shorter duration than their peers. Model 2 showed that both maternal education at the college level and the high school diploma level were positively associated with duration; similar to the identification models, race/ethnicity were no longer significant predictors after controlling for maternal educational attainment. Overall, these findings suggest that student characteristics play a significant role not only in predicting whether or not a student ever participates in a gifted program but also in  how long they stay, which may alter the impact of the program and offer insight into whether they had a positive or negative experience.

Further analysis of the relationship between gifted program participation, race/ethnicity, and socioeconomic status revealed that racial disparities also differ across levels of maternal education. As displayed in Figure 1, racial disparities in gifted identification are most prominent for students whose mothers have attended college. In this category, the proportion of Black students and non-Black/Hispanic students identified as gifted surpasses the proportion of Hispanic students by a mean difference of −0.071 and −0.078 respectively. In contrast, racial disparities in gifted participation duration is most evident for students whose parents do not hold a high school diploma. As displayed in Figure 2, both Black students and Hispanic students linger behind their peers in this category. These findings suggest that the parent characteristics that influence initial gifted identification may be different than the parent characteristics that influence gifted program persistence and/or how early a student is identified.

Overall, these findings prove that academic ability alone does not determine participation in gifted and talented programs. Participation and duration of participation are both associated with race/ethnicity, socioeconomic status, and gender, suggesting that a student’s background may introduce barriers to access. Underrepresented populations who already linger behind their peers on measures of academic achievement, also disproportionately receive gifted services, potentially contributing to social identity-based gaps in academic performance.

Academic Outcomes of Gifted Students

To explore the relationship between participation in a gifted and talented program and long term academic outcomes, I ran a regression analysis on five main outcomes variables at age 14: PIAT achievement test scores in math, reading recognition, and reading comprehension, academic self-concept, and educational aspirations. [4] As found in previous studies, gifted education has a clear positive effect on long term academic outcomes, as displayed in Table 3. Models 1, 3, 5, 7, and 9 display estimated effects for the entire sample, and Models 2, 4, 6, 8, and 10 display estimated effects for matched data.

Gifted students generally had greater academic outcomes than non-gifted students. On average, those who participated in a gifted program had math scores that were 8.9 points higher (out of 100), reading recognition scores that were 6.4 points higher, and reading comprehension scores that were 5.8 points higher than students who did not participate. Participation in a gifted program during elementary school years was also positively associated with academic self-concept and educational aspirations after controlling for demographic variables both pre- and post-propensity matching. In addition, being Black and being Hispanic were negatively associated with achievement test scores and academic self-concept, in alignment with previous studies. However, the association between race and academic self-concept did not remain significant after propensity matching, suggesting that the difference in academic self-concept between Black and non-Black/Hispanic students is greater for the general population than for those likely to participate in a gifted program. Having a mother who attended college also had a significant positive association with all academic outcomes with coefficients on par with or surpassing coefficients for gifted education. These findings suggest that race and maternal education are both significant predictors of academic performance, even within the population of gifted children. Moreover, in some cases, race and maternal education are greater predictors of academic performance and academic attitudes than gifted status.

These findings suggest that race and maternal education are both significant predictors of academic performance, even within the population of gifted children.

Although gifted education appears to have a clear positive association with academic outcomes, there was also notable heterogeneity in treatment effects by race/ethnicity and maternal education. As shown in Figures 3 and 4, the relationship between gifted education and achievement test scores were similar across racial/ethnic categories and maternal education levels, however, varied widely for academic self-concept, as well as educational aspirations. For academic self-concept, gifted education had a medium effect size [5] for students whose mothers held a high school diploma (d = 0.57) or attended college (d = 0.43), but little to no significant effect on those whose mothers did not graduate from high school (d = −0.02), suggesting that the observed differences in academic self-concept may not hold true for the population of students with lower levels of maternal education.

Similarly, when broken down by maternal education, the estimated effect of gifted education on educational aspirations seems to be driven by those whose mothers’ highest level of education is a high school diploma, for which the effect size is moderate and positive (d = 0.46) compared to smaller effect sizes for those whose mothers attended college (d = 0.32) and seemingly negative effects for those whose mothers did not obtain a diploma (d = −0.42). Similar patterns exist for race: the effect of gifted education on educational aspirations was moderate for Non-Black/Hispanic students (d = 0.46) but very small for Black (d = 0.11) and Hispanic students (d = 0.23). These findings suggest that the positive association between gifted education and attitudes toward academics may be driven by outcomes for more privileged students, specifically Non-Black/Hispanic students and students whose mothers’ highest level of education surpasses a high school diploma.

Social-Emotional Outcomes of Gifted Students

To examine the relationship between gifted education and social-emotional outcomes during high school years, multiple regression models were created for four key outcome metrics: the Self-Perception Profile for Children (SPPC) self-worth score, the Rosenberg Self Esteem Scale, the Pearlin Mastery Scale, and The Center for Epidemiological Studies Depression Scale (CES-D). As with academic outcomes, the association between gifted participation and social-emotional outcomes was estimated before and after propensity matching, the results of which are displayed in Table 4. Models 1, 3, 5, and 7 display estimated effects for the entire sample, and Models 2, 4, 6, and 8 display estimated effects for matched data.

Generally, students who participate in a gifted and talented program have higher levels of self-esteem (a mean score 0.74 points higher on the Rosenberg Self Esteem Scale) and self-worth (a mean score 4.21 points higher on the Self-Perception Profile for Children), and report greater confidence in their abilities and perception of control over their lives. As indicated in Table 4, there are positive associations between gifted participation and self-worth, self-esteem, and mastery both pre- and post- propensity matching after controlling for demographic variables. Race was not associated with self-perception, however, being Black was positively associated with depression before matching but not after, suggesting that the difference in depression levels between Black students and non-Black students is wider among the general population than among those who are likely to enroll in gifted programs.

Higher maternal educational attainment was also positively associated with self-worth, self-esteem, and mastery, and negatively associated with depression before and after matching. Moreover, maternal education was the greatest predictor of self-worth both at the high school diploma and college degree attainment levels, followed by gender, then gifted education. In addition, the relationship between maternal education and mastery was less impactful after matching, suggesting that maternal education is not as strong a predictor of mastery for students who are already likely to enroll in a gifted program than for students generally, regardless of whether they actually enrolled. Lastly, boys reported higher senses of self-worth, self-esteem, and mastery, and lower levels of depression than girls across all models, in alignment with previous research on gender disparities in gifted education. Generally, these findings suggest that while gifted education is correlated with better social-emotional outcomes, demographic factors play a significant, sometimes larger, role in predicting outcomes as well.

As seen with academic outcomes, there was also notable heterogeneity in the association between gifted education and social-emotional outcomes by race and maternal education. As displayed in Figure 5, there generally seems to be a stronger relationship between gifted education and social-emotional outcomes for Black and Hispanic students than for Non-Black/Hispanic students. When comparing differences in mean outcomes by race, we see that gifted education has a notable effect size on self-worth for Black students (d = 0.299), but little to no effect for Non-Black/Hispanic students (d = 0.083) and Hispanic students (d = 0.032). Similarly, the effect size is moderate for the association between gifted education and mastery for Hispanic students (d = 0.560) and Black students (d = 0.339), but imperceptible for Non-Black/Hispanic students (d = 0.093). In alignment, the interaction between gifted education and being Black is a significant predictor of self-worth (b = 7.871, p < 0.10) and mastery (b =  0.773, p <0.10), as shown in Appendix C. The interaction between gifted education and being Hispanic is also a significant predictor of mastery after propensity matching (b = 1.473, p < 0.01; see Appendix C), supporting the conclusion that the relationship between gifted education and social-emotional outcomes varies by race.

As shown in Figure 6, there are also notable differences in the association between gifted program participation and social-emotional outcomes based on socioeconomic status. For students whose mothers hold a high school diploma or attended college, the relationship between gifted education and outcomes are similar, however, there seem to be distinct patterns for those whose mothers hold less than a high school diploma. With a portion of the variability is due to the smaller sample size, there is also sufficient evidence that gifted education impacts these students differently. First, gifted education has a moderate effect size on mastery for those with a maternal education of less than a high school diploma (d = 0.407), compared to smaller effect sizes for those with higher levels of education (d = 0.239; d = 0.211). Similarly, there is a larger difference in mean depression levels for students with a maternal education of less than a high school diploma (d = −0.511) than for those whose mothers hold a high school diploma (d = 0.210) or attended college (d = 0.042). Furthermore, although gifted education was not a significant predictor of depressive symptoms for the entire matched sample, it was a significant predictor (b = −3.04, p < 0.01; see Appendix C) after controlling for the interaction between gifted education and maternal education. The interaction itself was also a significant predictor of depression, after matching, at both the high school diploma (b = 3.623, p < 0.01) and attended college (b = 3.161, p < 0.01) levels of maternal education, as displayed in Appendix C. These findings suggest that in contrast to patterns seen with academic outcomes, the association between gifted education and social-emotional outcomes is primarily driven by students of color, specifically Black and Hispanic students, and students whose mothers have lower levels of education.

Discussion

This study sought to explore barriers to entry into gifted and talented programs for Black and Hispanic communities and low-socioeconomic status students, as well as potential racial and socioeconomic disparities in the outcomes for those in gifted programs. My first research question surrounding the extent to which demographic variables affect the probability of participation in gifted and talented programs yielded findings similar to those found in previous studies. Results indicated that both race/ethnicity and maternal education play a significant role in predicting gifted program participation as a binary, as well as how long students remain in gifted and talented programs. These findings affirm that under current gifted education policies, less privileged populations are excluded from opportunities for advanced programming at a disproportionate rate than their peers. Since one of the primary criticisms of gifted education is the issue of equity and the potential of widening the achievement gap between gifted and non-gifted students, the outcomes of gifted students should be viewed through the lens of these racial and socioeconomic disparities in access.

These findings affirm that under current gifted education policies, less privileged populations are excluded from opportunities for advanced programming at a disproportionate rate than their peers.

My second research question sought to determine the impact of gifted and talented programs on students’ academic and social-emotional development, with a focus on outcomes during high school years. I find that students who participate in gifted programs have higher achievement test scores in reading and math, report greater levels of academic self-concept, and have higher educational aspirations during high school years than those who do not. They also score higher on measures of self-worth, self-esteem, and mastery than nongifted students. [6] 

Lastly, my third research question honed in on differences in outcomes by race/ethnicity and socioeconomic status. I found that the differences in academic attitudes between gifted students and non-gifted students are greater for non-Black/Hispanic students and students with higher levels of maternal education than their counterparts. Differences in social-emotional outcomes between gifted and non-gifted students, specifically measures of self-worth and mastery are greater for Black and Hispanic students, as well as students with lower levels of maternal education. Generally, my findings support the conclusion that gifted programs may impact students of color and students of low socioeconomic status in different ways than they impact students from more privileged backgrounds. Understanding this distinction may help offer insight into the mechanisms of gifted and talented programs, and how to better equalize the experience of gifted students across different backgrounds through policy interventions.

While this study offers valuable insight into racial and socioeconomic disparities in gifted education, there were several key limitations that restrict the generalizability of my findings. First, the present study relies on observational data, which is more prone to bias and confounding due to unobserved covariates. Although the study aimed to limit confounding through propensity matching, there were many factors not included in the analysis that could influence the relationship between gifted program participation, demographic factors, and long-term outcomes. More specifically, unobserved characteristics like grit, parent marital status, and zip code would make students more or less likely to be placed into gifted programs. Since these variables were not included in the data, they may not be evenly distributed across the matched samples of gifted and non-gifted students, leading to potential selection bias and omitted variable bias in analyzing student outcomes. Second, both a strength and limitation of this study is the utilization of a national longitudinal dataset as opposed to a more specific case study. While using nationally representative data offers more macro-level insight on gifted education than previous cross-sectional studies on specific schools, districts, and programs, it also limits the practical interpretation of my findings. Since there is wide variability in gifted identification policies and gifted program structure across states, the average findings presented in this study at the national level may not hold true at the state or local level. Future research might explore these themes and any further heterogeneity in the treatment effects of gifted education on a narrower scale that allows for potential causal claims and tailored policy recommendations, perhaps through a regression discontinuity study of students who were on the borderline of gifted identification thresholds.

Implications for Education Policy and Practice

The United States has made strides in reducing the racial achievement gap in K-12 education over the last few decades but there is still more progress to be made in addressing disparities in the classroom, especially for students of color and low-income students. Ability grouping is only one of the ways that schools inadvertently stratify students by race, ethnicity, socioeconomic status, and gender, contributing to existing inequities and boosting the outcomes of already high-achieving students. By failing to recognize giftedness and excluding underprivileged students from opportunities for advanced programming, like gifted education, magnet schools, or AP/IB classes, these policies restrict students from reaching their full academic potential. In alignment with prior literature, my findings support the following recommendations for state-level gifted and talented program policies:

1. Use holistic strategies of gifted identification that do not rely on achievement test scores or teacher recommendations. 

   Currently, the two most commonly used methods of gifted identification favor White students and students of higher socioeconomic status. In order to increase racial and socioeconomic diversity in gifted and talented programs, state and district-level program coordinators must invest in more comprehensive and inclusive admissions practices. Only about half of the states in the U.S. have acknowledged inequities in their gifted identification processes and have developed inclusion policies for increasing the identification of gifted minority students (McClain and Pfeiffer 2012, 72-75). These improvements include: allowing for flexibility in test score thresholds, incorporating more holistic forms of assessment like creativity tests and teacher ratings of specific characteristics (ex. motivation) instead of a binary recommendation of “gifted” or “not gifted” to reduce bias, and having school psychologists play a more central role in assessing students and advocating for those who are overlooked.

   One concrete example of alternative assessment can be drawn from the state of Iowa which considers portfolios of work, intake interviews, and “evidence of gifted characteristics”. Gifted program coordinators in Iowa also adjust test score metrics for minority and low-income students (McClain and Pfeiffer 2012, 72). Although employing these alternative forms of gifted identification for every student is infeasible due to financial restraints and time-intensive demands, one strategy for reducing costs is the use of a lower-quality and low-cost baseline assessment for all students, followed by a higher quality, multiple criteria assessment process (ex. achievement test, creativity test, portfolio, behavioral ratings, etc) for those who perform well on the initial screening. This way, all students have the opportunity to be identified without the barrier of teacher nomination, but are ultimately assessed on more equitable, yet methodologically sound instruments for evaluating giftedness.

2. Conduct a review of the experiences of minority students within gifted and talented programs, not only those applying, through an equity lens to determine how to best create a supportive learning environment for gifted students from disadvantaged backgrounds.

   Diversity, equity, and inclusion efforts should not end after the identification process. These findings demonstrate that academic and socioemotional outcomes are also predicted by race/ethnicity, income, and gender, suggesting that students of different social identities are receiving different levels of academic support in gifted classrooms. Further research must be conducted to fully understand these disparities and the connection between student experiences and student outcomes. Based on prior literature, shifts in classroom culture— such as celebrations of all students’ academic milestones— and increased education and support for the parents and caregivers of gifted students may help ease negative experiences. However, qualitative data collection and transparency are the first steps toward policy solutions. Legislation such as New Jersey’s Strengthening Gifted and Talented Education Act, which increases state oversight of district-run gifted and talented programs and requires districts to regularly collect and report data to the state, acts as foundational policy for this increased level of data availability.

3. Boost the emphasis on multicultural and culturally responsive pedagogy in teacher training programs, and recruit more gifted educators of color.

   Because gifted education programs attempt to capture a wide range of skills beyond those of traditional teacher education programs, it is easy for multicultural education to slip through the cracks. It is the responsibility of state-level gifted and talented program coordinators to advocate for baseline standards for teacher preparation programs in their state, specifically concerning inclusive instruction and facilitation. Based on prior literature, policy adjustments could include: increased training in multicultural education and culturally responsive pedagogy for gifted and talented teachers and/or greater incentives and recruitment efforts for teachers of diverse backgrounds. 

   Successful recruitment efforts that have been conducted throughout the United States to increase gifted and talented teacher diversity include:

   - increased student teaching and internship opportunities in gifted and AP classrooms,

   - district-provided financial incentives to encourage teachers of color to acquire gifted education credentials, and

   - partnership between school districts and Historically Black Colleges and Universities (HBCUs) to prepare and recruit Black teachers.

4. Launch school-based and/or district-level intervention programs for gifted minority students and gifted low-income students, i.e. those who qualify for free or reduced-price  lunch (FRPL).

   Beyond long term reform efforts directly in the classroom, one strategy that can be implemented in the short run is launching and funding school-based intervention programs that specifically target gifted students from underrepresented backgrounds. These programs often include services like out-of-school enrichment activities, parent support and education, tutoring, peer mentoring, cultural enrichment, and psychological services (Woo et. al 2017, 203), and have been proven to boost academic outcomes among gifted minority students. Project EXCITE, an intervention program facilitated by Northwestern University in partnership with two public school districts in Evanston, Illinois identified talented Black and Hispanic elementary students and front loaded STEM enrichment, as well as mentorship and parent support, resulting in significant long term improvement. The academic performance of participating students came close to the performance levels of white, Asian and non-low-income students (Olszewski-Kubilius and Steenbergen-Hu 2017, 206). State-funded interventions like these facilitated at the district level could be a promising framework for providing additional support and increasing retention for gifted minority students.

While ability grouping is not inherently egregious, it can lead to long-lasting disparities and perpetuate harmful misconceptions when lines are drawn by demographic factors, rather than student abilities and learning styles. It is the responsibility of educators and policymakers to ensure that ability grouping is implemented in a way that enhances the student experience, rather than setting limitations, and that the talents of all students are recognized, regardless of what they look like and where they come from.

*This article was by Auri Minaya (Princeton University), Rhythm Banerjee (Graduate Institute of Geneva), and Lea Hunter (Princeton University).

---

 About the Author

Born and raised in Jersey City, New Jersey, Krystal Cohen is a first-year MPA student at the Princeton School of Public and International Affairs with a focus on addressing racial, gender, and socioeconomic disparities in K-12 education. She received her BA in Sociology from Princeton University with certificates in Statistics and Machine Learning, and African American Studies. Most recently, she’s worked in fundraising and community outreach at Foundation Academy Charter School in Trenton, NJ.

(Return to Top)

---

Acknowledgements

Krystal would like to thank Professor Jennifer Jennings for her mentorship and guidance through the research process, the JPIA editing team for their thoughtful recommendations, and all those who supported her advocacy work throughout the years. 
 

---

Notes

1. Although there is quite a range of strategies for grouping gifted students, the term ‘gifted and talented programs’ typically refer to withdrawal programs, which have proven to be the most effective in influencing student outcomes.  (Return to Note)

2. Note: The term “gifted” here refers to students enrolled in gifted programs, not academic ability. “Non-gifted” refers to students of similar academic ability not placed in gifted programs.(Return to Note)

3. Due to the large number of observations missing age 5 test scores (N = 997), the dummy variable indicating missingness was included to preserve the sample size and allow for fair comparison of coefficients across models. (Return to Note)

4. Age 14 was chosen both to allow for effective comparison across birth years and as a mark of when the student likely transitioned to high school after receiving gifted education services throughout elementary school. (Return to Note)

5. Effect sizes as measured by Cohen’s d are interpreted based on guidelines set forth in “Estimating the Size of Treatment Effects” by James J. McGough and Stephen V. Faraone (2009). The term “effect size” does not refer to causal effects. (Return to Note)

6. In this context, the term “non-gifted” refers to students who did not participate in gifted and talented programs and is not related to cognitive ability. (Return to Note)

---

References

Anderson, Brittany N. 2020. “‘See Me, See Us’: Understanding the Intersections and Continued Marginalization of Adolescent Gifted Black Girls in U.S. Classrooms.” Gifted Child Today 43(2):86–100. doi: 10.1177/1076217519898216.

Bate, Joanne, and Deb Clark. 2013. “Like Minds Learning Well Together: Improving Academic, Social, and Emotional Outcomes for Gifted Students.” Set: Research Information for Teachers (1). doi: 10.18296/set.0350.

Bui, Sa A., Steven G. Craig, and Scott A. Imberman. 2014. “Is Gifted Education a Bright Idea? Assessing the Impact of Gifted and Talented Programs on Students.” American Economic Journal: Economic Policy 6(3):30–62.

Crawford, Brittany F., Kate E. Snyder, and Jill L. Adelson. 2020. “Exploring Obstacles Faced by Gifted Minority Students through Bronfenbrenner’s Bioecological Systems Theory.” High Ability Studies 31(1):43–74. doi: 10.1080/13598139.2019.1568231.

Dehejia, Rajeev, and Sadek Wahba. 2002. “Propensity Score-Matching Methods For Nonexperimental Causal Studies.” The Review of Economics and Statistics 84(1):151–61.

Delcourt, Marcia A. B., Dewey G. Cornell, and Marc D. Goldberg. 2007. “Cognitive and Affective Learning Outcomes of Gifted Elementary School Students.” Gifted Child Quarterly 51(4):359–81. doi: 10.1177/0016986207306320.

Ford, Donna Y., Tarek C. Grantham, and Gilman W. Whiting. 2008a. “Another Look at the Achievement Gap: Learning From the Experiences of Gifted Black Students.” Urban Education 43(2):216–39. doi: 10.1177/0042085907312344.

Ford, Donna Y., Tarek C. Grantham, and Gilman W. Whiting. 2008b. “Culturally and Linguistically Diverse Students in Gifted Education: Recruitment and Retention Issues.” Exceptional Children 74(3):289–306.

Grodsky, Eric, and John Warren. 2008. “Testing and Social Stratification in American Education.” Annual Review of Sociology 34. doi: 10.1146/annurev.soc.34.040507.134711.

Gross, Miraca U. M. 2006. “Exceptionally Gifted Children: Long-Term Outcomes of Academic Acceleration and Nonacceleration.” Journal for the Education of the Gifted 29(4):404–29. doi: 10.4219/jeg-2006-247.

Guzy, Annmarie. 2018. “Honors Is a Good Fit for Gifted Students-- Or Maybe Not.” Journal of the National Collegiate Honors Council 19(2):9–16.

Kulik, Chen-Lin C., and James A. Kulik. 1982. “Effects of Ability Grouping on Secondary School Students: A Meta-Analysis of Evaluation Findings.” American Educational Research Journal 19(3):415–28. doi: 10.3102/00028312019003415.

McBee, Matthew. 2010. “Examining the Probability of Identification for Gifted Programs for Students in Georgia Elementary Schools: A Multilevel Path Analysis Study.” Gifted Child Quarterly 54(4):283–97. doi: 10.1177/0016986210377927.

McBee, Matthew T., Scott J. Peters, and Erin M. Miller. 2016. “The Impact of the Nomination Stage on Gifted Program Identification: A Comprehensive Psychometric Analysis.” Gifted Child Quarterly 60(4):258–78. doi: 10.1177/0016986216656256.

McClain, Mary-Catherine, and Steven Pfeiffer. 2012. “Identification of Gifted Students in the United States Today: A Look at State Definitions, Policies, and Practices.” Journal of Applied School Psychology 28(1):59–88. doi: 10.1080/15377903.2012.643757.

McGough, James J., and Stephen V. Faraone. 2009. “Estimating the Size of Treatment Effects.” Psychiatry (Edgmont) 6(10):21–29.

Olszewski-Kubilius, Paula, and Saiying Steenbergen-Hu. 2017. “Blending Research-Based Practices and Practice-Embedded Research: Project Excite Closes Achievement and Excellence Gaps for Underrepresented Gifted Minority Students.” Gifted Child Quarterly 61(3):202–9. doi: 10.1177/0016986217701836.

Preckel, Franzis, Thomas Götz, and Anne Frenzel. 2010. “Ability Grouping of Gifted Students: Effects on Academic Self-Concept and Boredom.” British Journal of Educational Psychology 80(3):451–72. doi: 10.1348/000709909X480716.

Smith, Jocklyn, Barbara LeRose, and Robert E. Clasen. 1991. “Underrepresentation of Minority Students in Gifted Programs: Yes! It Matters!” Gifted Child Quarterly 35(2):81–83. doi: 10.1177/001698629103500206.

Stuart, Elizabeth A. 2010. “Matching Methods for Causal Inference: A Review and a Look Forward.” Statistical Science : A Review Journal of the Institute of Mathematical Statistics 25(1):1–21. doi: 10.1214/09-STS313.

Tomlinson, Carol Ann. 2000. “Differentiation of instruction in the elementary grades.” ERIC Digest. Syracuse, NY: Office of Educational Research and Improvement (ED), Washington, DC. 

U.S. Bureau of Labor Statistics. 2020. “National Longitudinal Survey of Youth 1979 Child and Young Adult | National Longitudinal Surveys.” Retrieved December 14, 2020 (https://www.nlsinfo.org/content/cohorts/nlsy79-children).

Vialle, Wilma, Patrick C. L. Heaven, and Joseph Ciarrochi. 2007. “On Being Gifted, but Sad and Misunderstood: Social, Emotional, and Academic Outcomes of Gifted Students in the Wollongong Youth Study.” Educational Research and Evaluation 13(6):569–86. doi: 10.1080/13803610701786046.

Wagner, Kathryn, Laurie Dymes, and Greg Wiggan. 2017. “Tracking Students Through Life: A Critical Structural Analysis of Academic Tracking of Mexican Immigrant Students in the United States and Korean Immigrant Students in Japan.” The Urban Review 49(5):875–94. doi: 10.1007/s11256-017-0425-x.

White, Gwyne W., Cesalie T. Stepney, Danielle Ryan Hatchimonji, Dominic C. Moceri, Arielle V. Linsky, Jazmin A. Reyes-Portillo, and Maurice J. Elias. 2016. “The Increasing Impact of Socioeconomics and Race on Standardized Academic Test Scores across Elementary, Middle, and High School.” American Journal of Orthopsychiatry 86(1):10–23. doi: 10.1037/ort0000122.

Woo, Hongryun, Na Mi Bang, Bridget Cauley, and Namok Choi. 2017. “A Meta-Analysis: School-Based Intervention Programs Targeting Psychosocial Factors for Gifted Racial/Ethnic Minority Students.” Journal for the Education of the Gifted 40(3): 199–219. Doi: 10.1177/0162353217717034

Zeidner, Moshe. 2020. “‘Don’t Worry—Be Happy’: The Sad State of Happiness Research in Gifted Students.” High Ability Studies 0(0):1–18. doi: 10.1080/13598139.2020.1733392.

 

---

Appendix

​​​

​