Assessing the Influence of No Child Left Behind on Gifted Education Funding in Texas: A Descriptive Study

Abstract

No Child Left Behind (NCLB) has defined the past 15 years of public K-12 education. An incentive structure built around adequate yearly progress created an environment that was not aligned with gifted education. Texas, with over 11% of the total identified gifted population in the United States, state funding for gifted, and incentivized identification policies, made an ideal case study to analyze the ramifications of NCLB on gifted education. This article explores how Texas responded to NCLB and that response’s influence on district-level funding for gifted education. In total, 16 years of financial and enrollment data were analyzed for the 1,025 public school districts in Texas using the frame work of a longitudinal mixed model. Results indicated that there was an annual decline in the percentage of budget allocated to gifted education of 0.04 percentage points for rural school districts, 0.08 for suburban, 0.07 for town, and 0.05 for urban.

Keywords

No Child Left Behind gifted education urban rural suburban longitudinal

Upon inception of No Child Left Behind (NCLB; U.S. Department of Education, 2003), scholars in the field of gifted education worried that the mandates in the law would have unintended negative consequences for gifted education (Gallagher, 2004; Gentry, 2006). Gallagher (2004) stated that the pressures of standardized testing would create teachers who focused on students passing tests rather than conceptually learning material. Gentry (2006) further pointed out that standardized tests lead to standardized curriculum focused on teaching to the middle. Kaplan (2004) stated that NCLB was an opportunity for the field of gifted education to reform. The tenants of increased accountability, focusing on the achievement gap, and creating standards were all philosophically aligned with best practices in gifted education. All authors agreed that the potential for focusing solely on minimal proficiency (through standardization) and creating strong incentives to promote minimal proficiency (through punitive measures based on standardized test scores) has the severe potential to hedge out excellence.

One possible consequence of NCLB on gifted education that scholars did not directly discuss was the reallocation of school resources away from gifted education. Understanding how money is spent and personnel are allocated in a school district can provide context to observed educational outcomes. That being said, education policy, including fiscal policy, has been woefully underexamined in gifted education research (Plucker, Makel, Matthews, Peters, & Rambo-Hernandez, 2017). Prior to Kettler, Russell, and Puryear’s (2015) examination of how demographics of a school district influence gifted education spending, the last comprehensive analysis of funding policy for gifted education was by Baker in 2001. In short, the entire policy life cycle of NCLB was unexplored by scholars in the field. This article seeks to rectify that gap within the literature.

This article provides a brief overview of NCLB, the funding structure for gifted education in Texas, and the influence of NCLB in Texas public education. Following this, a descriptive overview of Texas funding for gifted education programming is examined before and after the enactment of NCLB. Finally, a discussion of the descriptive analysis and future research directions are presented.

NCLB

NCLB lasted from 2001 to 2015 when it was replaced by the Every Student Succeeds Act (ESSA; Klein, 2016). The effects of NCLB on different facets of education are beginning to be explored by scholars. Dee and Jacobs (2011) analyzed National Assessment of Education Progress data between 1992 and 2007. The authors reported an overall increase of 7.2 scale points (effect size of 0.23) in mathematics performance for fourth-grade students following the enactment of NCLB. Furthermore, the authors also found a statistically significant positive difference for students performing in the top 10%. From the enactment of NCLB, the authors found a 5.205 point scale increase on the math portion of the National Assessment of Educational Progress scores for students in the top 10% of scores. One important factor found by the authors is that there was a difference between states that had accountability measures before the enactment of NCLB. Those states that had strong accountability measures before NCLB did not see the same rise in performance as states that did not. In the case of Texas, Dee and Jacobs (2011) noted that Texas had strong accountability measures and strong repercussions.

Furthermore, Dee, Jacobs, and Schwartz (2013) noted that per pupil spending increased by US$600 during the time frame of NCLB. The authors examined the Common Core of Data’s Local Education Agency Finance Survey and the School and Staffing Survey between the 1994-1995 and 2007-2008 academic schools years. The authors found that the number of teachers with advanced degrees increased as well as teacher salaries. The authors did note that teachers reported shifting time away from science and social studies to focus on tested subjects (reading more so than math). The authors posture that the achievement-based incentives caused district administrators to shift financial resources to “teacher and pupil support services” (Dee et al., 2013).

Jennings and Lauen (2016) examined the effect of NCLB on school educator behavior in Houston Independent School District (ISD) in Texas (the seventh largest school district in the United States). They found that in schools that were struggling to meet the adequate yearly progress (AYP) mandates of NCLB had large performance gaps between the state mandated test and an audit test (the Stanford Achievement Test [SAT]). What the authors claimed, through their results, is that schools in Houston ISD that were more at risk of not meeting AYP were more likely to teach to the test (as shown by the gap between the state and audit test). They found that this gap was almost double for Black children in comparison with Hispanic children.

When faced with difficult decisions related to resource allocation, superintendents must decide what programs can be cut. McNamara (2015) interviewed two superintendents in Illinois who were facing budgetary shortfalls. The superintendents stated that during budget reductions, discretionary programs needed to be reduced. The discretionary area that was cut was gifted programs.

In summation, scholars have demonstrated that NCLB achieved its targets in terms of increasing student performance (Dee & Jacob, 2011) and in increasing qualifications of the national teaching corps (Dee et al., 2013). Unfortunately, the legislation also had unintended consequences. The incentive structure around AYP encouraged educators to teach students test mastery rather than content mastery (Jennings & Lauen, 2016). Furthermore, it created an environment wherein administrators made difficult choices in terms of resource allocations that were likely not with the best interests of gifted programming in mind (McNamara, 2015).

NCLB and Funding

The federal government was able to leverage these mandates on states through potentially withholding Title 1 funds from noncompliant states (Ladd, 2017). Despite only accounting for 1.5% of total spending on education, the threat of withholding Title 1 funds led to compliance in all states (Ladd, 2017). To meet the mandates of NCLB, states increased per pupil spending by US$570 from the enactment of the act in 2001 to the 2007-2008 school year (Dee & Jacob, 2010). Of the increase in funding, US$470 of the US$570 went to hiring teachers and increasing teacher salaries. The remaining increase in funding went to educational support services (e.g., administration). In contrast, Dee and Jacobs (2010) noted that there was no corresponding increase in federal funding. In essence, states, and in turn school district administrators, had to meet the increased demands leveled by NCLB without additional resources provided by the federal government. It is important to note, though, that NCLB and other policy are not the only factors when considering funding for gifted education.

The Influence of Locale on Gifted Education Budgets

Kettler et al. (2015) examined the effect of locale on gifted education funding in the state of Texas. The authors used a cross-sectional approach to analyze the 2010-2011 fiscal year. The authors discovered that rural school districts allocated fewer resources in terms of personnel and money to gifted education programming despite having the greatest property value per student in Texas (US$328,295). In contrast to rural school districts, Kettler et al. (2015) found that the suburban districts allocated the greatest resources (in terms of proportion of budget and personnel) to gifted education despite having the lowest per pupil property values (US$263,786). An important distinction in understanding this apparent contradiction is that the property wealth for rural school districts is primarily commercial and residential for suburban districts. Howley, Rhodes, and Beall (2009) noted that the combination of declining populations and poor property wealth can create severe funding shortage in rural schools. Fewer students means less per pupil revenue, coupled with declining property values (due to less property demand on already weak property markets), can force superintendent to shift priorities away from gifted education programming.

In contrast to rural schools, Walberg and Fowler (1987) noted that large urban districts have funding problems due to their large size. Large districts, spread over numerous schools, with large personnel staff, can lead to inefficient allocations of resources. Van Tassel-Baska (2010) noted that urban school districts are often in need of greater per pupil resources than their suburban counterparts. Despite this, Kettler et al. (2015) did not find a significant difference in the funding of urban schools compared with suburban districts in their cross-sectional analysis of the 2010-2011 Texas fiscal year.

The Influence of Ethnicity on Gifted Education Budgets

The direct relationship between funding for gifted education and the ethnic composition of a district has been woefully understudied. Kettler et al. (2015) did not find a significant relationship between the ethnic composition of a district and how that district funds and staffs its gifted education programs in Texas. Baker (2001) stated that, in comparison with the United States at large, Texas had an strong funding structure which supported its diverse population.

When scholars have directly discussed ethnicity and gifted education budgets, it has been in the context of adequacy (Moon, 2009; Renzulli, 2011). Moon (2009), in a discussion on the myth that high ability students do not experience difficulties, noted that students from traditionally underrepresented populations need additional resources and support to succeed. In line with this, Renzulli (2011) noted that specialized programming and identification procedures should be implemented to serve students from traditionally underrepresented population in his discussion on defining giftedness. In other words, it is not that ethnic compositions of districts directly influence budgetary decisions made for gifted programming, but that budgetary decisions shape and influence the underrepresentation in gifted programs (Ford, 2014).

Texas School Funding Structure for Gifted Education

The funding structure in Texas is founded upon the idea that all students should be equally funded. The Texas legislature has translated this ethic into a system of reappropriations. As stated by the Texas Education Agency (TEA), in Texas, schools receive income from property taxes of residents and business within the school district’s municipal borders. Under this scheme, certain districts are deemed to be property rich by the state and others as property poor (TEA, 2013). The established benchmark set by the state legislature to determine property wealth status is the school district of the state capital city, Austin ISD (TEA, 2013). If a district has more property wealth per student than Austin ISD, any money over that threshold is taken by the state and redistributed to poorer districts.

Under this system, the funding and identification of gifted students is incentivized by the state. Funding for gifted education is part of the base entitlement for a school district (TEA, 2013). School districts receive money for each student identified as gifted up to 5% of the total enrollment of the district. What this means is that for property rich districts, they are able to keep more of their money if they identify 5% of their student population. For property poor districts, they are entitled to more money if they identify 5% of students as gifted. This incentive structure has led to standardized rates of identification clustered around 5% across state public schools (TEA, 2013).

The caveat to this funding structure is that only 55% of funding received for gifted programming must be used for gifted programming. The remaining 45% are at the discretion of the district to be used on indirect costs (TEA, 2013). This was enacted to allow district administrators flexibility in responding to individual school district population needs. It is important to note that the definition of what an indirect cost entails is very broad. An indirect cost can describe funding for a school psychologist or for utilities for a building. In these cases, a school psychologist administers identification tests for the district and so indirectly supports the gifted program. Furthermore, without the electricity bill being paid, having school would be difficult. These examples illustrate the level of discretion that a district has in determining what an indirect cost might be. Superintendents in Texas would face discretionary decisions with the mandates of NCLB in Texas.

NCLB in Texas

When the United States adopted NCLB, the nation adopted the education accountability system of Texas (Heilig & Darling-Hammond, 2008). President George W. Bush took the educational accountability system he oversaw as governor of Texas and instituted it on a nationwide scale. Heilig and Darling-Hammond (2008) chronicled how Texas legislators sought to not only meet the standards set by their former governor but be an exemplar state through increasing already in place accountability measures.

Prior to NCLB, Texas had accountability measures in education (Dee & Jacob, 2011; Heilig & Darling-Hammond, 2008). In 1980, Texas implemented the Texas Assessment of Basic Skills (TABS) test (Confrey & Carrejo, 2002). In 1984, the TABS was replaced by the Texas Assessment of Minimal Skills (TEAMS) test. From 1991 to 2002, Texas school districts were evaluated based on their performance on the Texas Assessment of Academic Skills (TAAS). With the passing of NCLB, the TAAS test was replaced by the Texas Assessment of Knowledge and Skills (TAKS) test (Heilig & Darling-Hammond, 2008).

Although Texas had a 20-year history of accountability prior to NCLB, Texas still altered its accountability structure. Heilig and Darling-Hammond (2002) noted three key areas where Texas shifted accountability policy: (a) the minimum percentage of students dropping out was lowered from 6% to 5%, (b) an alternative test was developed to include special education students and limited English proficiency students in accountability measures, and (c) a social studies subject test was added, increasing the breadth of testing.

Compliance with NCLB standards were not immediately enforced. States were allowed certain time frames to meet the enforced standards. For example, states were given until the 2005-2006 academic year for their teaching staff to attain “highly qualified” status. Texas achieved a 95% rate in the 2002-2003 year (Texas House of Representatives, 2004). In the following academic year, after relaxation of qualification criteria for rural teachers, Texas was able to meet this mandate (Texas House of Representatives, 2004). In contrast, AYP went into effect immediately in Texas (Texas House of Representatives, 2004). Prior to NCLB, Texas had an accountability measure in the state (Dee & Jacob, 2011; Heilig & Darling-Hammond, 2008). NCLB increased those standards which led to an increase in schools not meeting standards across the state (Texas House of Representatives, 2004).

Heilig and Darling-Hammond (2008) noted that the increased accountability measures led to a “gaming” attitude among administrators. The authors use the example of exit exams required to graduate. Administrators were able to “game” the system by having low-performing students held back in the ninth grade. Following this, they were promoted to an at-age level grade. In this way, these students were not used to assess 10th-grade performance but instead were only counted in the 11th- and 12th-grade exit exam takers. Doing so removed the student from the pool of test-takers that the school was accountable for (10th graders) and put the student in a pool of test-takers where accountability was less high stakes (11th and 12th graders).

Purpose

The purpose of this article is to descriptively assess the influence of NCLB on budgetary allocations toward gifted education in the state of Texas. Texas mandates gifted services and provides funding for those services. Texas is ideal to examine how policy affects gifted education spending because districts in Texas are not obligated to use all of their available funding for gifted education on gifted education. This allows a researcher to assess how the increasing pressures of standardization can influence budgetary allocations. Before researchers can fully understand the influence of NCLB on student outcomes, a foundation of what transpired after the implementation of NCLB must first be understood. The purpose of this research is to provide one possible link through time series modeling of school district budgets prior to NCLB and of the years following.

A further purpose is to increase the body of literature associated with gifted education policy. In a recent call to action, Plucker et al. (2017) stated the need for increased policy research in gifted education. The authors decried the fact that policy research about gifted education has often been pursued by researchers outside the field of gifted education (e.g., economists). In response to this, the authors called on researchers in gifted education to more closely assess the effectiveness and influence of policy related to gifted education.

Understanding budgetary changes in gifted education funding allows researchers to better frame the consequences of policy decisions. Resource and personnel allocations provide the link between policy and outcomes. The purpose of examining budgetary changes is to provide possible answers to why outcomes are affected by policy changes. In short, to fully address the call to policy research made by Plucker et al. (2017), the intermediaries between policy and outcomes must first be examined. In the case of this research, the intermediary to be examined is the influence of NCLB on funding. Understanding that influence will help frame the consequences of NCLB policy on gifted education outcomes.

Research Questions

Research Question 1: To what extent did the enactment of NCLB influence direct funding for gifted programming in the state of Texas?

Research Question 2: To what extent was the influence of NCLB on direct funding related to school district characteristics?

Method

Data were acquired by public request from the TEA and through the TEA website. The dataset included all school districts in Texas between the 1999-2000 and 2014-2015 academic school years. Charter school districts receive funding differently than public school districts and were removed from the dataset. The dataset included 16 years of budgetary and enrollment data for the 1,025 public school districts in Texas, encompassing the period before NCLB and up to its replacement by ESSA. The structure of the dataset is nested with repeated annual budget and demographics under school districts.

The financial dataset included all budgetary items, revenue, and expenditures, for school districts. As such, all sources of revenue for a school district are included in the dataset. Furthermore, all expenditures for school districts are included in the dataset. In total, there are 245 financial variables available for analysis (TEA, 2013). The demographic dataset included only the number of students by race and the number of identified gifted students by race for each school district in Texas by year. These two datasets were merged using R 3.40 (R Core Team, 2017).

Model Statement

The following model was assessed:

\begin{array}{l} Y_{i t} = α + β_{1} {(preNCLB)}_{t} + β_{2} {(NCLB)}_{t} + β_{3} {(preNCLB)}_{t}^{2} + β_{4} {(NCLB)}_{t}^{2} \\ + β_{5} {(Proportion Black)}_{i t} + β_{6} {(Proportion Hispanic)}_{i t} \\ + β_{7} {(Proportion Native)}_{i t} + β_{8} {(Rural)}_{i} + β_{9} {(Suburban)}_{i} + β_{10} {(Urban)}_{i} \\ + β_{11} {(District revenue)}_{i t} + β_{12} {(Percentage identified as gifted)}_{i t} \\ + β_{13} {(Proportion Black)}_{i t} \times {(NCLB)}_{t} + β_{14} {(Proportion Hispanic)}_{i t} \\ \times {(NCLB)}_{t} + β_{15} {(Proportion Native)}_{i t} \times {(NCLB)}_{t} + β_{16} {(Rural)}_{i} \\ \times {(NCLB)}_{t} + β_{17} {(Suburban)}_{i} \times {(NCLB)}_{t} + β_{18} {(Urban)}_{i} \times {(NCLB)}_{t} \\ + [μ_{0 i} + μ_{1 i} {(preNCLB)}_{i t} + μ_{2 i} {(NCLB)}_{i t} + μ_{3 i} {(Proportion Black)}_{i t} \\ + μ_{4 i} {(Proportion Hispanic)}_{i t} + μ_{5 i} {(Proportion Native)}_{i t} \\ + μ_{6 i} {(District revenue)}_{i t} + μ_{7 i} {(Percentage identified as gifted)}_{i t}] + e_{i t} . \end{array}

This equation describes that the percentage of budget allocated to gifted education in year t by school district i is predicted by the year, the rate of year to year change, the proportion of traditionally underrepresented students in year t, locale, the school district’s overall revenue in year t, the percentage of students identified as gifted in the district in year t, the interaction between the underrepresented student variable and the years following NCLB, and the interaction between locale and the years following NCLB. Furthermore, to account for the nested structure of the data, the effect of year, proportion of traditionally underrepresented students, district revenue, and the percentage of students identified as gifted are allowed to vary by school district.

All associated R code used in this analysis can be provided upon request to the corresponding author. Associated R code includes data preprocessing and analysis allowing a reader to fully replicate the analysis in this article. Given the size of the dataset and complexity of the model, computational run time could be extensive. Furthermore, for readers using HLM 7, a decomposed model statement can be provided upon request to facilitate replication.

Furthermore, associated datafiles for district budgets are located on the TEA website. Demographic datafiles were provided by public data request to TEA but can be provided to readers upon request to the corresponding author. Datafiles are offered in raw form so that a reader may evaluate all the associated code, including the code for data preparation.

Description of Variables

Dependent variable

The dependent variable used in the analysis was the percentage of budget allocated to gifted education in a district. Texas districts are mandated to report their budgetary spending to the state on an annual basis. Percentage of budget is a useful metric to discuss budgetary changes as it is robust to inflation. The percentage of budget allocated to gifted education only entails direct funds used for gifted education programming.

Time variables

The primary independent variables of interest in the model are the time variables. A spline was coded for the time period before NCLB and one for the years following. The 3 years prior (preNCLB) were coded in consecutive order (i.e., 1999-2000 was 0, 2000-2001 was 1, 2001-2002 was 2; all remaining years were coded as 2). The second spline was coded in a similar fashion to the first. The years 2002-2003 to 2014-2015 (NCLB) were coded consecutively starting at 1 (e.g., 2002-2003 as 1, 2003-2004 as 2, etc.) and the years 1999-2000, 2000-2001, and 2001-2002 were coded as 0 (see Table 1 for an illustration of the coding scheme used). Time was coded in this manner to assess the relationship between an increase in one unit of time (1 academic year) and the dependent variable. Thus, this variable can be interpreted as the annual change in budgetary allocations to gifted education programming. To find the rate of change in the time slope, the square of the time variables “preNCLB” and “NCLB” were included in the model referred to as “preNCLB annual change” and “NCLB annual change.”

Table 1.

Coding Scheme for the Time Variables.

Year	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	2011	2012	2013	2014
PreNCLB	0	1	2	2	2	2	2	2	2	2	2	2	2	2	2	2
NCLB	0	0	0	1	2	3	4	5	6	7	8	9	10	11	12	13

Note. NCLB = No Child Left Behind.

Independent variables

Two sets of independent variables and two covariates were added to the model. The first set of independent variables was the proportion of Black, Hispanic, or Native American students in the school district. The second set controlled for differences in locale as defined by the National Center for Education Statistics (NCES; 2017). A covariate was added that controlled for difference in budgetary scale between districts. Finally, a covariate was added to the model that controlled for the percentage of gifted students, centered at 5%.

Proportion of Black, Hispanic, and Native American students

Kettler et al. (2015) used the variable non-White in their cross-sectional analysis. In line with this, a set of three variables were included that described the proportion of Black, Hispanic, and Native American students in a school district. A set of three variables were calculated by dividing the total number of Black, Hispanic, or Native American students by the total school population in the same year. One variable was calculated each for Black, Hispanic, and Native American students. Unlike Kettler et al. (2015), Asian students were excluded. Asian students are consistently overrepresented in gifted populations in comparison with Black, Hispanic, and Native American students (Yoon & Gentry, 2009). In this way, only the proportion of traditionally underrepresented students is considered. This is a time varying variable.

The inclusion of race within the model serves as an extension of the Kettler et al. (2015) findings. The authors found no link between the racial composition of a district and funding for gifted education. In this model, the influence of race and funding will be examined in the context of a major policy change.

Locale

The difference in budgetary allocations for gifted programs by locales in Texas was observed by Kettler et al. (2015). To control for the difference in funding between locales, school districts were coded using the NCES locale codes. These codes are derived from an area’s population density and distance from an urban cluster. A set of four dummy variables were coded for each locale (rural, suburban, town, and urban) where 1 signified the locale. For the model to be nonrank deficient, the locale dummy variable for town was dropped from the model and used as baseline instead. This is not a time varying variable.

Furthermore, the NCES provides 12 possible locale codes. There are three locale codes for rural, suburban, town, and urban locales. These codes are based on distance from urban centers (rural, suburban, town) or size of urban center (urban). To reduce model complexity and increase interpretability, these codes were aggregated into their superordinate codes (rural, suburban, town, urban).

Given the findings of Kettler et al. (2015), the inclusion of locale variables is necessary as funding for gifted is demonstrably linked to locale. Furthermore, their inclusion allows for an extension of the work of Kettler et al. (2015) by assessing the influence of locale during a major policy change.

Standardized total revenue

District budgets varied greatly. In Texas, there are school districts with budgets over a billion dollars (e.g., Houston ISD) and districts in rural Texas with budgets under a million dollars. Walberg and Fowler (1987) noted that economic scale could influence the efficiency of schools. To control for economic scale, a standardized financial variable was included in the model. In other words, total district revenue was rescaled to have a mean of zero and a standard deviation of 1. This means that each unit increase in the regression is equivalent to 1 standard deviation increase. This is a time varying variable. Furthermore, this variable was standardized within year.

Gifted

Funding in Texas is capped at 5% of student population. It is probable that this cap has influenced identification rates in the state. To control for this influence, the percentage of gifted students in a given year for a school district was included in the model. Furthermore, this variable was centered on 5%. As such, this means that the intercept can be interpreted as the mean percentage of budget allocated to gifted education at the maximum level of state funding when the coefficient for gifted percentage is 0 in the regression. Finally, this variable was scaled such that a unit increase in the regression is equivalent to a 1% increase over the 5% cap. This is a time varying variable.

Model fitting and analysis

All statistical analysis in this research was done using R 3.40 (R Core Team, 2017) and the lme4 package for R (Bates, Mächler, Bolker, & Walker, 2014). Furthermore, lme4 uses restricted maximum likelihood (REML) as default for estimation over maximum likelihood (ML). Given that no model comparisons are being conducted and that j is sufficiently large (j = 1,025), REML is an appropriate estimator (Bates et al., 2014).

To assess model fit, an unconditional model was constructed using only time variables. The Bayesian information criterion (BIC) was assessed when examining the wellness of fit between different time models. The time models tested were linear time and quadratic time. To assess the appropriateness of a mixed model approach, the intraclass correlation was examined. Finally, the model coefficient of determination was calculated in the manner described by Xu (2003). Xu (2003) stated that traditional means of calculated model coefficients of determination (r²) could lead to incorrect estimations and suggested an alternative approach. Xu proposed the use of $Ω^{2}$ as a pseudo-r². This pseudo-r² is calculated by subtracting from one the proportion of residual variance of the full model to the residual variance of the null model.

Finally, as this is an exploratory analysis and descriptive study that encompasses an entire population, the use of p values is inappropriate (Greenland et al., 2016). For readers who are uncomfortable with possible measurement error due to district reporting of financial/demographic information or who wish to extend the model to predict future values in the state of Texas, R code can be provided upon request that will calculate associated p values. The lmerTest package provides p values for coefficients derived from lme4 (Kuznetsova, Brockhoff, & Christensen, 2015).

Results

A means table was constructed to provide an initial examination of the dependent variable (see Table 2). Furthermore, a correlation table of the dependent variable with the independent variables was created (see Table 3). Finally, two figures were created to provide a visual representation of the dependent variable over time. The first was the overall means plot of the dependent variable (see Figure 1). A second figure was created to separate means by locale (see Figure 2).

Table 2.

Percentage of Budget Allocated to Gifted Programs Between the 1999-2000 and the 2014-2015 Academic School Year.

Academic school year	Number of school districts	Percentage of budget allocated to gifted programs (SE)
1999-2000	1,025	1.34 (0.04)
2000-2001	1,025	1.35 (0.04)
2001-2002	1,025	1.35 (0.04)
2002-2003	1,025	1.31 (0.04)
2003-2004	1,025	1.24 (0.04)
2004-2005	1,025	1.18 (0.03)
2005-2006	1,025	1.16 (0.03)
2006-2007	1,025	1.11 (0.03)
2007-2008	1,025	1.06 (0.03)
2008-2009	1,025	1.00 (0.03)
2009-2010	1,025	0.98 (0.03)
2010-2011	1,025	0.96 (0.03)
2011-2012	1,025	0.98 (0.03)
2012-2013	1,025	0.90 (0.03)
2013-2014	1,025	0.88 (0.03)
2014-2015	1,024	0.86 (0.03)

Table 3.

Correlation Matrix for Independent Variables.

Variable	1	2	3	4	5	6	7	8	9	10	11	12
1. NCLB	1
2. PreNCLB	.18	1
3. NCLB Annual Change	−.96	−.15	1
4. PreNCLB2 Annual Change	−.24	−.99	.20	1
5. Rural	−.03	−.01	<.01	<.01	1
6. Suburban	.02	<.01	<.01	<.01	.53	1
7. Urban	.08	.01	.01	<.01	.28	.29	1
8. % Gifted	.11	<.01	−.05	.01	.02	.01	.01	1
9. % Black	.01	<.01	.04	<.01	.12	.02	−.03	.02	1
10. % Hispanic	−.12	−.01	−.02	<.01	.09	.02	−.05	.01	.19	1
11. % Native	−.04	−.01	.06	.01	.01	−.01	<.01	.02	.01	.06	1
12. Revenue	−.07	<.01	.04	<.01	.01	−.12	−.29	−.09	−.15	−.13	−.02	1

Note. Values listed as <.01 denote values between .0049 and −0.0049. NCLB = No Child Left Behind.

Figure 1.

The percentage of budget allocated to gifted programs in the state of Texas between the 1999-2000 and the 2014-2015 academic school year with error bars included.

Figure 2.

The percentage of budget allocated to gifted programs in the state of Texas between the 1999-2000 and the 2014-2015 academic school year by NCES locale (rural, suburban, town, and urban).

Diagnostics

An initial examination of the intraclass correlation provided evidence that a mixed model was appropriate (intraclass correlation coefficient [ICC] = .65). This was expected considering that the study involved repeated measures of school districts over time. A means plot was created and assessed to provide a basis for initial fitting of the longitudinal model (see Figure 1). The BIC was analyzed for best model fit for time variables. A linear fit and linear fit with quadratic terms were examined for the years following NCLB. A linear fit with quadratic terms (BIC = 60,432.82) proved to be more appropriate than a linear fit (BIC = 60,449.41). This was confirmed more formally by an ANOVA (F = 26.92, p < .001). A table showing the selected unconditional model can be seen in Table 4.

Table 4.

Regression Results.

Fixed effects	Unconditional model		Full model
Fixed effects	Beta	t ^a	Beta (SE)	t ^a
Intercept	1.42 (0.09)	16.05	1.69 (0.11)	14.87
NCLB	−0.06 (<0.01)	12.85	−0.07 (0.01)	8.06
PreNCLB	−0.09 (0.09)	1.01	−0.09 (0.07)	1.37
NCLB Annual Change	<0.01 (<0.01)	5.19	<0.01 (<0.01)	7.07
PreNCLB Annual Change	0.02 (0.02)	1.06	0.02 (0.02)	1.59
Rural			−0.61 (0.10)	6.37
Suburban			0.25 (0.11)	2.26
Urban			0.23 (0.22)	1.07
% Gifted			1.96 (0.27)	7.20
Black			−0.04 (0.30)	1.32
Hispanic			−0.16 (0.14)	1.22
Native			−1.30 (1.12)	1.15
Revenue			0.41 (0.14)	2.83
NCLB × Rural			0.03 (0.01)	3.53
NCLB × Suburban			−0.01 (0.01)	0.25
NCLB × Urban			0.02 (0.02)	1.00
NCLB × Black			−0.02 (0.03)	0.84
NCLB × Hispanic			−0.01 (0.01)	0.51
NCLB × Native			0.09 (0.17)	0.52
Random effects	Variance (SD)		Variance (SD)	Correlation^b
Intercept	1.18 (1.09)		1.60 (1.27)
NCLB			0.01 (0.11)	−.37
PreNCLB			0.22 (0.48)	−.55
Revenue			2.60 (1.61)	−.24
% Gifted			6.80 (2.61)	.48
Black			1.03 (1.01)	−.10
Hispanic			1.52 (1.23)	.27
Native			0.94 (0.97)	−.76
Residual	0.56 (0.75)		0.28 (0.53)
Fit statistics	Value		Value
BIC	60,432.82		48,543.12
Ω²	.68		.85

Note. NCLB = No Child Left Behind; BIC = Bayesian information criterion.

The Wald t value is provided to provide clarity on the ratio of beta coefficients and their associated standard error.

This variable is the correlation between the intercept and random effect.

Model assumptions for linear mixed models were examined as well using R. The QQ-plot of the fixed effect residual error provided evidence for a roughly normal distribution, though with some evidence for heavy tails. Considering that the sample size is sufficiently large (n > 1,000), the presence of heavy tails is unlikely to bias estimates (Faraway, 2014). The QQ-plot for random effect residual error provided evidence of light tails. Again, though, considering that sample size is sufficiently large, the presence of light tails is unlikely to bias estimates (Faraway, 2014). The residual plot provided evidence for homogeneity. Furthermore, a residual versus fitted plot demonstrated that the assumption of linearity was upheld as no visible pattern was detected.

Finally, the BIC for the conditional model was 48,620.18. The model coefficient of determination, $Ω^{2}$ , was 85.44.

Model Results

Full model results can be seen in Table 4. Results from the model suggested that rural school districts allocated 0.61 percentage points (SE = 0.10) less of their budget to gifted program than town and 0.86 and 0.84 percentage points less than suburban and urban locales, respectively. This provides evidence that on average, rural school districts allotted 1.08 (SE = 0.10) percentage points of their budget to gifted education program. In contrast, suburban districts allocated 1.94 percentage points (SE = 0.11), towns allocated 1.69 percentage points (SE = 0.11), and urban districts allocated 1.92 (SE = 0.22) percentage points.

Model results indicate that there has been decline in the percentage of budget allocated to gifted education after the enactment of NCLB. This decline was a 0.04 (SE = 0.01) percentage points for school districts in rural locales, 0.08 (SE = 0.01) percentage points for suburban locales, 0.07 (SE = 0.01) percentage points for town locales, and 0.05 (SE = 0.02) percentage points for urban locales. To provide an example of the magnitude of these coefficients, the baseline locale, town, will be used as an example. The coefficient for town (the intercept) indicates that over a 10-year period, there was a decline in budget allocations by 0.70 percentage points. When the average school district in Texas allocated 1.31% of its budget to gifted education in the 2002-2003 academic school year, a decline of 0.70 percentage points is not a trivial decline. Finally, the beta estimates for suburban and urban locales should be interpreted with caution due to their large associated standard error.

It should be noted that there was a less than 0.01 (SE < 0.01) percentage point slowdown in the annual rate of the decline across the state. Prior to the enactment of NCLB, there was a decline in budgetary allocations of 0.09 percentage points but the associated standard error of 0.07 suggests that caution should be used when assessing the certainty of this estimate. These results largely align with the means plot shown in Figure 1.

The percentage of gifted in a school district had a beta coefficient of 1.96 (SE = 0.27). Furthermore, the proportion of Black students was associated with a 0.04 percentage point (SE = 0.30) reduction in percentage of budget allocated to gifted education. The proportion of Hispanic students was associated with a 0.16 percentage point (SE = 0.14) decline in percentage of budget allocated to gifted education. The proportion of Native American students was associated with a 1.30 percentage point (SE = 1.12) decline in percentage of budget allocated to gifted education. Finally, there was a 0.41 percentage point (SE = 0.14) change in percentage of budget allocated to gifted education for every standard deviation from the mean state school district revenue.

In terms of random effects, the variance component of the intercept suggests that budgetary allocations for gifted education are not uniform across the state (SD = 1.27). Furthermore, the variance component for NCLB demonstrates that the budgetary decline was not uniform across districts. Furthermore, the –.37 correlation with the intercept suggests that school districts with greater budgetary allocations to gifted education had greater rates of decline.

Finally, an examination of Figure 2 confirms the coefficients obtained during the exploratory regression analysis. Clearly, rural districts allocate a smaller proportion of their budget to gifted education. Furthermore, the year to year decline is the less steep of the four locale types. For suburban districts, there appears to be a lag after the enactment of NCLB and the decline of budget allocations to gifted education. With the exception of the lag, the budgetary reduction appears to be similar to urban districts. Finally, town districts appear to have a similar decline to suburban and urban districts but allocate less of the budgets to gifted education. The degree to which they allocate less compared with suburban and urban districts does not appear to change over time.

Discussion

The results from this article provide statistical support to the predictions of Gallagher (2004) and Gentry (2006) that the enactment of NCLB disincentivized funding for gifted education in Texas public schools. Although there is a statistical correlation, it does not provide evidence of true causality. That being stated, the funding structure for Texas public schools had been enacted in 1996 and remains unchanged (TEA, 2013). Statistically, there is an average annual percentage point decline in budget allocations of 0.04 for rural school districts, 0.08 for suburban school districts, 0.07 for town school districts, and 0.05 for urban school districts. To put this into context, when the average suburban school district budget allocation to gifted education is 1.94% of their total average school district budget, a 5-year change of 0.08 percentage points is more than a quarter reduction in budget. This does not necessarily mean that budgets were cut directly and immediately following NCLB. As the results indicate, the decline in relative funding was gradual.

The results here provide quantitative evidence to the qualitative evidence found by McNamara (2015). The author interviewed two superintendents and found that in the face of crisis, gifted education was an easy cut to make. These results provide evidence that the findings of McNamara are not isolated but can be generalized, at the very least, into the state of Texas.

In general, there was a shift in the value structure of public education due to the incentives structured around NCLB. As Dee et al. (2013) and Jennings and Lauen (2016), AYP was a powerful incentive tool that influenced the decisions of administrators and educators. NCLB created an atmosphere that rewarded teaching to the test. This reward structured largely did not favor gifted education programming.

In essence, the predictions stated by Gallagher (2004) and Gentry (2006) have largely come to pass in the state of Texas, at least in the realm of financial allocations. Caution should be taken, though, in drawing the conclusion to adequacy of programming and financial allocations to that program. The funding of a program is not perfectly correlated with the quality or outcomes of that program. For example, Dee and Jacobs (2011) provided evidence for increase in math scores of students in the top 10% of test-takers 5 years after the enactment of NCLB on a national level. Regardless, the findings from Dee et al. (2013) and Jennings and Lauen (2016) strongly provide evidence to the shift in focus to standardized testing predicted by Gentry (2006).

In other words, money talks. Where money is allocated is a good indication of what is valued. In the case of gifted education, the focus on AYP has led district administrators to view gifted education as a luxury.

There has been a slowdown in the annual rate of decline of budgetary allocations to gifted education. Although this is at first promising, in Texas, there is a mandatory minimum for gifted education spending. At least 55% of that funding for gifted programs provided by the state must go directly to gifted education programs. In other words, what this means, in line with the results from the analysis, is that less of the discretionary funding gifted education is being spent on gifted education. Instead, these funds are likely being allocated toward indirect costs (e.g., central office salaries or M&O costs). The funding incentives from Texas incentivizes districts to continue to identify and maintain gifted programs (TEA, 2013), but NCLB has created a second set of incentives which has led to a siphoning of potential funds from gifted programs.

Furthermore, the variance component for the random effect of NCLB provides evidence that the decline was not uniform across districts. Furthermore, a negative correlation of –.37 means that those districts that allocated greater proportions of their budgets to gifted education were more adversely affected. For every standard deviation increase in a school allocated budget, the coefficient for NCLB decreased by .04. Put in another way, for schools that allocated 1.6% of their budget to gifted education, these schools experienced an annual decline of 0.07 percentage points. In contrast, those schools that allocated 2.87% of their budget to gifted education (1 standard deviation above the mean) experienced a decline of 0.11 percentage points. This relationship is unsurprising as schools with greater budgetary allocations toward gifted education likely had more of their gifted funding allocated toward direct funds rather than indirect.

With regard to the second research question, rural school districts had a lower annual rate of decline than other district locales. Again, this might seem heartening, but must be put into context of other statistical results. In contrast, the standard errors for the point estimates for urban and suburban districts suggest that making a claim about their annual rate of decline is less certain. Given this and the visual evidence from Figure 2, there was likely little to no difference between the annual rate of decline between suburban and urban districts with town districts.

Regardless, rural districts had lower rates of average decline than all three of the other locale types. In reality, though, this is likely due to rural districts having lower overall levels of funding compared with other locales. A coefficient of −0.61 for rural implies that an intercept of 1.69 means that rural districts allocate 1.08% of their budgets to gifted compared with the mean. What this result means, especially given the visual evidence in Figure 2, is that rural districts were likely operating at levels much closer to the state mandated minimum for their gifted programs before the enactment of NCLB than other locales. In other words, a possible interpretation to the fact that rural districts have a lower rate of decline than other school districts is that their initial operating budgets at the onset of NCLB were lower rather than active measures taken in rural districts.

This finding confirms and extends the work of Kettler et al. (2015) who also found that rural districts allocate fewer resources and personnel to gifted education. This finding extends their work by showing that their finding is not isolated to a single year, but has been an ongoing trend, and one that is independent of NCLB. In short, it is not that gifted programs in rural schools were not adversely affected by NCLB, it is that the ongoing economic and social causes cited by Howley et al. (2009) supersede NCLB.

Furthermore, the findings suggest that that proportion of students from traditionally underrepresented groups (Black, Hispanic, and Native American) in a district has little bearing on how much of its budget is allocated to gifted education. This, again, extends the findings of Kettler et al. (2015). Furthermore, it extends the findings by providing the context of NCLB. Even with the pressure of NCLB, those districts with higher proportions of underrepresented students did not make more extensive resource allocations away from gifted programming than other districts. It has been over 15 years since Baker (2001) gave his assessment of Texas finance and gave an endorsement of Texas’s financial scheme in terms of equity. These results provide evidence that the observations made by Baker still hold true.

The findings presented here provide evidence to the efficacy of Texas’s financial policy for funding gifted education. NCLB created incentives to redistribute funds for gifted programs into other avenues. The discretionary funds were meant as “good-faith” funds by legislators to allow districts to better serve the needs of those districts’ unique student populations (TEA, 2013). These funds were envisioned to be allocated to sources that would benefit students identified for gifted services. This article provides strong evidence that, in fact, those discretionary funds were systematically allocated away from gifted education programs to other areas in the district. For example, a possible scenario is that districts used the discretionary funds for gifted programming to pay for building operations costs (e.g., electricity and waste management). These costs must be paid to keep the school operating but how a school district distributes its available funds provides an insight to the priorities of that school district.

Furthermore, Texas already had accountability measures prior to the inception of NCLB (Dee & Jacob, 2011). This means that, in a state that already had accountability measures, the additional accountability measures from NCLB created an environment where the incentives toward standardization were so strong, that resources were still funneled from gifted programs. If the state did not have a system of accountability in place, leading to a culture of accountability, the transition to such a new paradigm could at least partially explain the difference in allocation of budget.

One positive must be acknowledged: the funding strategy of a mandatory minimum is an effective means of maintaining funding in school districts. Although this seems obvious, it is important for a reader to understand the policy implications of a mandatory minimum. As 55% of funding is mandatory and cannot be spent on indirect supports, Texas provides funding to districts at the base level of 2.75% of students identified for gifted services (TEA, 2013). This floor means that even if new incentive structures were enacted, there would always be a base level of funding for gifted programming in the state. Although Gallagher (2004) and Gentry (2006) correctly predicted the shift of resources that happened in Texas, the state legislatures, at bear minimum, created a “fail-safe” that would ensure gifted programs would be funded at some level.

Limitations

The first limitation is in regard to generalization of the findings. This study is only comprised of a single state, despite the state’s size. As different states have different policies, it is unlikely that the results can be generalized to states with disparate policies.

A second limitation is the truncated time frame prior to the enactment of NCLB. There were only three time points prior. Although this is the minimum needed to fit a time slope, the analysis made would have been strengthened by additional data points prior to the enactment of NCLB. There is a possibility that the time slope was not flat prior to NCLB. Texas does not maintain electronic records of its school districts’ finances prior to the 1999-2000 academic school year.

A third limitation is in the quantification of locale. Puryear and Kettler (2017) noted that the use of proximity in the NCES definition created instances where locales were not distinct from one another. For example, the authors found that rural locales in close proximity to urban clusters were characteristically different to rural locales further from urban clusters. The use of the NCES definition was chosen to quantify locale to further generalizability between studies. Despite this, the limitation of these definitions must be acknowledged.

Future Research

The fact that budgets have shrunk leads to the question of what are the consequences of shrinking budgets. To fully explore the consequences of NCLB on outcomes, the intermediary policy decisions should first be understood. For example, identification of underrepresented students in a district might decline. The direct cause is not NCLB but the fact that budgetary shrinkage led to alternative identification measures being discarded due to cost or personnel being shifted to different position. This research establishes a direct consequence of NCLB: budgets were shrunk. In this light, future research should examine how NCLB has indirectly affected education outcomes. The aforementioned representation is one facet but there is also college entrance, SAT scores, participation in dual credit/AP courses, or even degree attainment rate.

Conclusion

The work here constitutes a clear contribution to the call made by Plucker et al. (2017) through examining the intersection of state financial policy and federal policy and the effect on gifted education spending. As Plucker et al. (2017) stated, the analysis of gifted education policy should be done by gifted education researchers. Policy research has been woefully under researched in the field. This research aims to meet the call issued by Plucker et al. (2017) by examining state-level gifted education funding policy. The United States is a patchwork of different fiscal, identification, and programmatic policies. In essence, each state is its own self-contained policy experiment. The issue is that the results from these experiments have yet to be fully analyzed and understood.

As such, this research has focused on how the funding policy in Texas was affected by an overarching national-level policy. Results provide policy makers and researchers with evidence that large mandates can shift discretionary funds away from gifted education. The question then is posed if this is acceptable or if it is short-sighted. With this awareness, policy makers can make more informed decisions.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

About the Author

Jaret Hodges is a postdoctoral associate at the Talent Identification Program at Duke University. His research interests are in underrepresented populations in gifted education.

References

Baker

B. D.

(2001). Living on the edges of state school-funding policies: The plight of at-risk, limited-English-proficient, and gifted children. Educational Policy, 15, 699-723.

Bates

Mächler

Bolker

Walker

(2014). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67, 1-48.

Confrey

Carrejo

(2002). A content analysis of exit level mathematics on the Texas Assessment of Academic Skills: Addressing the issue of instructional decision-making in Texas. Retrieved from https://eric.ed.gov/?id=ED471756

Dee

T. S.

Jacob

B. A.

(2010). The impact of No Child Left Behind on students, teachers, and schools. Brookings Papers on Economic Activity, 2, 149-194.

Dee

T. S.

Jacob

B. A.

(2011). The impact of No Child Left Behind on student achievement. Journal of Policy Analysis and Management, 30, 418-446.

Dee

T. S.

Jacob

B. A.

Schwartz

N. L.

(2013). The effects of NCLB on school resources and practices. Educational Evaluation and Policy Analysis, 35, 252-279.

Faraway

J. J.

(2014). Linear models with R. Boca Raton, FL: CRC press.

Ford

D. Y.

(2014). Segregation and the underrepresentation of Blacks and Hispanics in gifted education: Social inequality and deficit paradigms. Roeper Review, 36, 143-154.

Gallagher

J. J.

(2004). No child left behind and gifted education. Roeper Review, 26, 121-123.

10.

Gentry

(2006). No child left behind: Neglecting excellence. Roeper Review, 29, 24-27.

11.

Greenland

Senn

S. J.

Rothman

K. J.

Carlin

J. B.

Poole

Goodman

S. N.

Altman

D. G.

(2016). Statistical tests, P values, confidence intervals, and power: A guide to misinterpretations. European Journal of Epidemiology, 31, 337-350.

12.

Heilig

J. V.

Darling-Hammond

(2008). Accountability Texas-style: The progress and learning of urban minority students in a high-stakes testing context. Educational Evaluation and Policy Analysis, 30, 75-110.

13.

Howley

Rhodes

Beall

. (2009). Challenges facing rural schools: Implications for gifted students. Journal for the Education of the Gifted, 32(4), 515-536.

14.

Jennings

J. L.

Lauen

D. L.

(2016). Accountability, inequality, and achievement: The effects of the No Child Left Behind act on multiple measures of student learning. RSF: The Russell Sage Foundation Journal of the Social Sciences, 2, 220-241.

15.

Kaplan

S. N.

(2004). Where we stand determines the answers to the question: Can the no child left behind legislation be beneficial to gifted students? Roeper Review, 26, 124-125.

16.

Kettler

Russell

Puryear

J. S.

(2015). Inequitable access to gifted education: Variance in funding and staffing based on locale and contextual school variables. Journal for the Education of the Gifted, 38, 99-117.

17.

Klein

. (2016, March 31). The Every Student Succeeds Act: An ESSA overview. Education Week. Retrieved from http://www.edweek.org/ew/issues/every-student-succeeds-act/index.html?cmp=eml-enl-eu-news1

18.

Kuznetsova

Brockhoff

P. B.

Christensen

R. H. B

. (2015). lmerTest: Tests for random and fixed effects for linear mixed effect models (lmer objects of lme4 package). R package version 2.0-6. 2014. R-project.org/package=lmerTest.

19.

Ladd

H. F.

(2017). No child left behind: A deeply flawed federal policy. Journal of Policy Analysis and Management, 36, 461-469.

20.

McNamara

E. A

. (2015). The effects of the 2008 financial recession on superintendents’ leadership and decision making (Doctoral dissertation). Retrieved from eCommons, Loyola University Chicago (Accession No. 1288).

21.

Moon

S. M.

(2009). Myth 15: High-ability students don’t face problems and challenges. The Gifted Child Quarterly, 53, 274-276.

22.

National Center for Education Statistics. (2017). Common core of data: Identification of rural locales. Retrieved from http://nces.ed.gov/ccd/rural_locales.asp#justification

23.

Plucker

J. A.

Makel

M. C.

Matthews

M. S.

Peters

S. J.

Rambo-Hernandez

K. E.

(2017). Blazing new trails: Strengthening policy research in gifted education. Gifted Child Quarterly, 61(3), 210-218.

24.

Puryear

J. S.

Kettler

(2017). Rural gifted education and the effect of proximity. Gifted Child Quarterly, 61, 143-152.

25.

R Core Team (2017). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org/

26.

Renzulli

J. S.

(2011). What makes giftedness? Reexamining a definition: Giftedness needs to be redefined to include three elements: Above-average Intelligence, high levels of task commitment, and high levels of creativity. Phi Delta Kappan, 92, 81-88.

27.

Texas Education Agency. (2013). School finance 101: Funding of Texas public schools. Retrieved from http://tea.texas.gov/WorkArea/DownloadAsset.aspx?id=2147511834

28.

Texas House of Representative Committee on Public Education. (2004). Interim report 2004. Retrieved from http://www.house.state.tx.us/_media/pdf/committees/reports/78interim/public_education.pdf

29.

U.S. Department of Education. (2003). Preliminary overview of programs and changes included in the No Child Left Behind Act of 2001. Retrieved from http://www.ed.gov/nclb/overview/intro/progsum/sum_pg2.html

30.

Van Tassel-Baska

. (2010). The history of urban gifted education. Gifted Child Today, 33(4), 18-27.

31.

Walberg

H. J.

Fowler

W. J.

Jr. (1987). Expenditure and size efficiencies of public school districts. Educational Researcher, 16, 5-13.

32.

(2003). Measuring explained variation in linear mixed effects models. Statistics in Medicine, 22, 3527-3541.

33.

Yoon

S. Y.

Gentry

. (2009). Racial and ethnic representation in gifted programs: Current status of and implications for gifted Asian American students. Gifted Child Quarterly, 53(2), 121-136.