(0.57)
Note . Estimates correspond to means and those in brackets correspond to standard deviations.
During the spring of kindergarten, school administrators were given a survey that included a question about whether students were required to wear uniforms. Although similar questions were asked again in first and second grade, these surveys were: (a) only asked of administrators of students who switched to new schools (and only a handful of children changed between schools with and without uniforms) and (b) no uniform data were available for students who switched schools after second grade, hence our focus on students who remained in the same elementary schools between kindergarten and fifth grade. Accordingly, we used these reports from kindergarten to create a binary variable that indicated whether students attended elementary schools with or without a uniform policy.
In the fall and spring of kindergarten and again in the spring of each subsequent grade, teachers reported on students’ socio-emotional skills. These questions were derived from the Social Skills Rating System (SSRS; Gresham & Elliott, 1990 ). This tool is based a 4-point scale (0 = never to 3 = very often ) that includes four subscales: Interpersonal skills (5 items; α = .86–.87), self-control (4 items; α =.80–.82), internalizing behavior problems (4 items; α = .76–.79), and externalizing behavior problems (5 items; α = .86–89). Teachers also reported on students’ approaches to learning with a measure developed by NCES (6 items; α = .91–.92). Similar to Claessens and colleagues (2009), we collapsed these indicators into three dimensions: Internalizing behavior problems , externalizing behavior problems , and social skills (a combination of approaches to learning and socioemotional skills).
In addition to children’s socio-emotional development, students’ teachers also reported on students’ school absences every year (0 = no absences , 1= 1–4 absences , 2 = 5–7 absences , 3 = 8–10 absences , 4 = 11–19 absences , and 5 = 20 or more absences ). To increase interpretability, we recoded the scale values to equal the midpoint of the response options (e.g., 1–4 absences was recoded as 2.5 absences). Students who were never absent (scale value of 0) and those who were absent for 20 or more days of the school year (scale value of 5) were coded as being absent for 0 and 20 days, respectively. Note that, during kindergarten through third grade, children had one primary teacher across subject areas who reported on children’s school attendance. In fourth and fifth grade, however, students had different teachers for different subject areas. In these grades, both students’ English language arts teacher and their science or math teacher responded to questions of absenteeism. Because the correlations of absenteeism across subject areas were high, we created a composite of fourth, and then, fifth grade absences.
All analyses were estimated within Stata ( StataCorp, 2009 ). These models included robust standard errors to safeguard against violations of normality and missing data were accounted for with 50 imputed datasets using chained equations. All models were also weighted to be nationally representative and error term were clustered at the school level. To minimize the possibility of spurious associations, all models controlled for a large number of child and family covariates. These indicators capture children’s characteristics (i.e., gender, race/ethnicity, English language learner status, and an indicator for whether a parent rated the child as having poor health), children’s educational experiences (i.e., enrollment in full-day kindergarten, school type, and the number of hours that the child spent in center-based prekindergarten and before/after school care during the kindergarten year), household characteristics (i.e., household structure, number of siblings, poverty status, parent education, parent employment, number of books in the home, home learning activities), and school-going practices and routines (i.e., whether the child took a school bus to school, how far the child lived from school, in miles, number of breakfasts and dinners that the family regularly had together at home). In addition, all models (except for models predicting absenteeism) adjusted for lagged dependent variables from kindergarten entry. Given the large number of outcomes, we also make a p -value adjustment for multiple comparisons using the Benjamini adjustment (Benjamini & Hochberg, 1995).
With the above analytic framework in mind, we employed several different methodological specifications to determine whether students demonstrated different outcomes based on school uniform policies. Our first model was based within an OLS regression framework. Importantly, the above is after taking into account a student’s own school entry skills and the control measures. Thus, our first set of analyses considered whether, conditional on covariates, students in schools with uniforms demonstrated greater improvements in outcomes between kindergarten and fifth grade as compared with students who attend schools without uniforms. Notwithstanding the rich control measures included in this study, it is important to note that schools with uniform policies may differ in other observed and unobserved ways, which would make it difficult to isolate the outcomes of school uniform policies from the effects of other factors. To limit this possibility, we estimated three additional models.
Our second specification addressed the possibility of variation at the state-level that may influence the associations between school uniform policies and student outcomes. We did so by implementing state-fixed effects for the full sample of children. Consequently, our state-fixed effects models hold constant all state-wide factors that were the same for students in schools with and without school uniform policies in the same state. Although state-fixed effects account for state-to-state variation, there may be variation at a more granular level in the implementation of school uniforms. Therefore, our third specification was based within a county fixed effects framework, which may be particularly important in a study of school uniform policies, as decisions about school uniforms can stem from county factors. Therefore, county fixed effects help to control for county-to-county variation that exists in school uniform policies and student outcomes. As before, both the state- and county-fixed effects models adjusted for lagged dependent variables and the full set of covariates discussed above.
Even though state- and county fixed effects account for observed and unobserved differences at key levels, and thus, increase confidence in the reported associations, there may be concerns about the overlap between schools with and without uniform policies. Accordingly, our fourth and final specification was based within a propensity score matching framework ( Rosenbaum & Rubin, 1983 ). Although propensity scores do not change the causal identification strategy, this methodology does consider whether there is overlap in the unmatched sample and the functional form assumptions that are driving our findings. For our matching models, we used the nearest neighbor method (with up to four matches) with a caliper of .05, allowing a sufficient overlap between students in schools with and without school uniforms. Given this specification, we successfully matched approximately 65% of students (the number of matches varied across the 50 imputed datasets). Importantly, before matching, the average standardized mean difference between conditions was approximately 18% of a standard deviation, but after matching, the average standardized mean difference was roughly 3% of a standard deviation (see Table 1 ). Moreover, none of the covariates were significantly different across conditions after matching, suggesting that balance was successfully achieved (descriptives available from authors). Accordingly, regression models were re-estimated within these matched samples and included all covariates when predicting outcomes (doubly robust estimation; Funk et al., 2011 ).
Once the main effects of school uniforms were examined in these various ways, we then examined potential variation in the benefits of school uniforms as a function of child and school characteristics. Specifically, we examined variation in associations as a function of child socioeconomic status and initial skills, and as a function of the type of school children attended. It is important to note that, given the small number of private schools of different types, we collapsed our school type indicator into public versus private for our moderation analyses. To estimate heterogeneity in associations, we estimated a new set of regression models that included interaction terms between the focal indicator for school uniforms and the moderators of interest. Our focal moderation analyses were estimated with the full set of covariates.
We begin with a descriptive presentation of the types of schools that had school uniform policies along with the students who attended those schools. We then present our main effect analyses before we turn to a discussion of heterogeneity in outcomes and close with set of supplemental analyses. With that said, and as can be seen in Table 1 , roughly 28% of students across the U.S. attended schools that required a uniform. When looking across different types of schools, we find that 78% of Catholic schools that students attended had a uniform requirement as compared with only 54% of other religious schools and 43% of other private schools. And, among public school students, only 21% attended schools with a uniform policy. In terms of the student body, we find that schools with uniforms served a larger number of Black (20%) and Hispanic (40%) children and English Language Learners (27%) than schools without school uniforms (8–18%). In contrast, White children were more likely to be served in schools without uniforms (64% vs. 31%), whereas schools with uniform policies served a larger share of children from low-income families (52%) than schools without uniform policies (41%). Other descriptives stratified by schools with and without uniform policies are presented in Table 1 .
Having established the descriptive snapshot of the schools with school uniform policies along with the student body, we next examined whether students demonstrated different outcomes based on school uniform policies. Two overall patterns are evident in Table 3 with regard to the associations between school uniforms and children’s social skills, behavioral problems, and school absences. First, the associations between school uniforms and students’ outcomes were almost entirely null and the effect sizes across outcomes and grade levels were roughly 3% of a standard deviation. In fact, of the 24 associations estimated within our baseline model with an assortment of covariates, only one emerged as statistically significant and none were statistically significant with a Benjamini adjustment, leaving us with little confidence of a statistically-significant link between school uniforms and students’ outcomes. Second, the magnitude of the estimated associations derived from our baseline OLS model did not change substantially when we estimated models with state- and county fixed effects, nor when we implemented propensity scores (see Table 3 ). Taken together, there seems to be no meaningful differences in students’ social-behavioral and attendance outcomes between kindergarten and fifth grade as a function of school uniforms.
Results from models examining the outcomes of school uniforms between kindergarten and fifth grade.
Student outcome | ||||
---|---|---|---|---|
Model specification | Social skills | Internalizing behavior | Externalizing behavior | Absenteeism |
Kindergarten outcomes | ||||
Covariate adjusted model | −0.02 (0.03) | 0.04 (0.04) | 0.04 (0.03) | −0.01 (0.05) |
State fixed effects model | −0.02 (0.04) | 0.03 (0.05) | 0.04 (0.03) | 0.02 (0.05) |
County fixed effects model | 0.00 (0.04) | 0.05 (0.05) | 0.02 (0.04) | 0.03 (0.05) |
Propensity score model | 0.00 (0.04) | −0.01 (0.04) | −0.00 (0.04) | −0.01 (0.05) |
First grade outcomes | ||||
Covariate adjusted model | −0.03 (0.05) | 0.04 (0.04) | 0.03 (0.04) | −0.03 (0.04) |
State fixed effects model | 0.01 (0.05) | −0.01 (0.05) | 0.02 (0.05) | −0.04 (0.05) |
County fixed effects model | 0.01 (0.05) | 0.01 (0.05) | 0.03 (0.05) | −0.02 (0.05) |
Propensity score model | 0.05 (0.05) | −0.03 (0.05) | −0.01 (0.05) | 0.01 (0.04) |
Second grade outcomes | ||||
Covariate adjusted model | −0.10 (0.05) | 0.04 (0.05) | 0.04 (0.04) | −0.04 (0.05) |
State fixed effects model | −0.09 (0.05) | 0.05 (0.05) | 0.02 (0.04) | −0.03 (0.05) |
County fixed effects model | −0.05 (0.05) | 0.00 (0.05) | 0.01 (0.04) | −0.03 (0.05) |
Propensity score model | −0.08 (0.05) | −0.01 (0.05) | 0.04 (0.05) | −0.02 (0.05) |
Third grade outcomes | ||||
Covariate adjusted model | −0.04 (0.05) | 0.04 (0.04) | 0.03 (0.04) | 0.01 (0.05) |
State fixed effects model | −0.00 (0.05) | 0.05 (0.05) | 0.00 (0.04) | 0.01 (0.05) |
County fixed effects model | 0.02 (0.05) | 0.03 (0.05) | −0.02 (0.04) | 0.01 (0.06) |
Propensity score model | −0.02 (0.05) | −0.02 (0.05) | 0.02 (0.04) | 0.02 (0.05) |
Fourth grade outcomes | ||||
Covariate adjusted model | −0.03 (0.04) | −0.02 (0.04) | 0.02 (0.04) | −0.01 (0.05) |
State fixed effects model | −0.03 (0.05) | 0.02 (0.05) | 0.00 (0.05) | −0.03 (0.05) |
County fixed effects model | 0.02 (0.04) | −0.02 (0.05) | −0.02 (0.04) | −0.01 (0.05) |
Propensity score model | −0.03 (0.04) | −0.05 (0.05) | 0.03 (0.04) | 0.01 (0.05) |
Fifth grade outcomes | ||||
Covariate adjusted model | −0.07 (0.05) | −0.02 (0.05) | 0.03 (0.05) | −0.01 (0.05) |
State fixed effects model | −0.02 (0.05) | −0.05 (0.05) | −0.01 (0.05) | −0.04 (0.05) |
County fixed effects model | 0.03 (0.05) | −0.08 (0.06) | −0.04 (0.05) | −0.03 (0.05) |
Propensity score model | −0.06 (0.05) | −0.04 (0.05) | 0.05 (0.05) | 0.01 (0.05) |
Note. Coefficients in bold were statistically significant at p < .05 with a Benjamini false discovery adjustment. All estimates are weighted and account for the complex sampling design. Models include a full set of controls. All continuous predictors and outcomes have been standardized to have a mean of 0 and standard deviation of 1 and, thus, coefficients can be interpreted as effect sizes. Estimates in brackets correspond to standard errors.
Having established the average associations between school uniforms and student outcomes, our next set of analyses examined heterogeneity in these associations as a function of children’s school entry skills, socioeconomic status, and the type of school students attended. In the main, there was no consistant evidence of variation in outcomes as a function of school type nor students’ baseline skills. There was, however, some indication that the links between school uniforms and absenteeism varied as a function of socio-economic status. More specifically, the attendance benefits of school uniforms were approximately 20% of a standard deviation larger for low-income students as compared with more affluent students. Accordingly, even though school uniforms had no links to attendance for higher-income students, in first ( p <.01), fourth ( p < .05), and fifth ( p < .05) grade, low-income students who attended schools with uniforms demonstrated fewer absences than those in schools without uniforms. And although not statistically significant, similar patterns emerged in second and third grade (ES = .10–.13)
In addition to the kindergarten through fifth grade outcomes reported as part of our focal analyses, students also reported on their school belonging (14 items, α = .90; e.g., closeness with teachers and classmates), experiences of bullying (4 items, α = .81; e.g., teasing, name calling), and social anxiety (3 items, α = .88; worrying about what others think) in fifth grade, which are aspects of the school experience that have been at the center of school uniform debates (e.g., Gentile & Imberman, 2011; Han, 2010 ; Howe, 1996 ; Huss, 2007 ; Kaiser, 1985 ; Pate, 2006 ; Sanchez et al., 2012 ). Although all students reported on these additional items, these surveys were not administered to students in the earlier grades. But to highlight the other potential outcomes of school uniform mandates, we estimate supplemental models that consider the links between school uniforms and these self-reported outcomes, net of the covariates outlined above. But because these variables were not collected in kindergarten entry, these models did not include lagged controls.
As can be seen in Table 4 , results from our covariate adjusted models revealed that students in schools with uniform policies reported no differences in their social anxiety, and experiences with victimization, but they did report lower levels of school belonging (ES = 16% of a standard deviation, p < .001) as compared with students in schools without school uniforms. These findings largely replicated when accounting for state- and county-fixed effects along with propensity scores (see Table 4 ) and remained statistically significant with a Benjamini adjustment. And, as before, there was no evidence of heterogeneity.
Results from models examining the associations between school uniforms and students’ self-reported outcomes in fifth grade
Student outcome in fifth grade | |||
---|---|---|---|
Model specification | School belonging | Experiences of bullying | Social anxiety |
Covariate adjusted model | − | 0.08 (0.04) | 0.03 (0.04) |
State fixed effects model | − | 0.09 (0.05) | 0.04 (0.05) |
County fixed effects model | − | 0.09 (0.05) | 0.05 (0.05) |
Propensity score model | − | 0.05 (0.04) | 0.03 (0.04) |
School uniform policies have grown in the thirty years since their introduction in the United States, both in the public and private education sectors ( U.S. Department of Education, National Center for Education Statistics, 2013 ). This growth has fostered debate among pro- and anti-uniform advocates ( Boutelle, 2008 ; Brown, 1998 ; Brunsma, 2006 ; DeMitchell, 2006 ; Kaiser, 1985 ; Pate, 2006 ; Portner, 1996 ; Ryan & Ryan, 1998 ; Zernike, 2002 ). To date, however, these debates have been grounded in both limited and dated empirical evidence, especially in the formative elementary school years ( Bodine, 2003 ; Brunsma, 2004 , 2006 ; Draa, 2005 ; Gentile & Imberman, 2011; Han, 2010 ; Kohn, 1998 ; Murray, 1997 ; Pate, 1999 ; Yeung, 2009 ).
With that said, Social Learning Theory posits that individuals rely on a combination of cognitive, behavioral, and environmental factors to learn how to act in a given situation. This process, known as reciprocal determinism, has been raised by policymakers and adminstrators to suggest that school uniform policies have downstream effects on students because it influences the environment of school children and alters the complex reciprocal determinism equation (Grusec, 1994). Accordingly, the present study sought to add to this literature on the going to school in the context of school uniforms in elementary schools by bringing a longitudinal and national perspective to school uniforms and aspects of students’ development that are relatively understudied, but that are known to drive long-term educational and life success ( Heckman & Rubinstein, 2001 ; Heckman et al., 2006 ). In doing so, several key themes emerged.
First, the results of the present investigation illustrate consistent and largely null findings at the aggregate level as a function of school uniform policies. That is, students who attended schools with and without school uniform mandates, on average, demonstrated similar social skills, externalizing and internalizing problems, and school attendance patterns between kindergarten and fifth grade after adjusting for children’s characteristics and their educational experiences, household characteristics, and school-going practices and routines. Effect size were close to zero, suggesting no meaningful differences as a function of school uniform policies. Importantly, this pattern of largely null findings remained consistent even with the inclusion of state and county fixed effects that accounted for geographic variation as well as when propensity score matching was used. But when looking at students’ own self-reports of their engagement and well-being in fifth grade, we found that students in schools with uniforms reported lower levels of school belonging than students in schools that did not require uniforms.
When taken together, these null—and in some cases negative—findings are both similar to prior studies that have documented null or negative associations ( Brunsma & Rockquemore, 1998 ; Gentile & Imberman, 2011; Han, 2010 ), but stand in contrast with other studies that show benefits of school uniforms for children’s socio-emotional development ( Huss, 2007 ; Wade & Stafford, 2003 ; Sanchez et al., 2012 ; Tucker, 1999 ). Although we can only speculate why these differences emerge between the current investigation and some of the extant literature, one must consider the fact that many of the prior studies done on school uniforms have been restricted to specific school districts (e.g., Draa, 2005 ; Gentile & Imberman, 2011; Hoffler-Riddick & Lassiter, 1996 ), whereas the current study presents a national perspective. But with regards to the Social Learning Theory perspective, there is no evidence to suggest school uniforms changed the environment of school children, at least with respect to their social behavior.
One might also wonder why there were largely null associations for broader indicators of social behavior (as reported on by teachers) as compared with negative associations for students’ own self-reports. Of most relevance is the fact that these benchmarks were different: Teachers reported on broader indicators of students’ social and behavioral adjustment, whereas students reported on more specific outcomes related to their school experiences. Thus, including indicators from both the teacher and student perspective presents a more well-rounded and balanced portrait of the outcomes of school uniform policies. But with respect to the lower levels of school belonging in schools with uniforms, one possibility worth considering is that students’ fashion choices are likely to be only one potential source of belonging. Accordingly, what these results make clear is that the argument that school uniforms create cohesion among students and give students a sense of identity is not true, at least in this study sample ( Brown, 1998 ; Kaiser, 1985 ; Pate, 2006 ; Ryan & Ryan, 1998 ; U.S. Department of Education, 1996 ).
The second key theme that emerged from the present investigation was that the magnitude of associations between elementary schools with (versus without) a school uniform mandate and students’ social and behavioral problems did not consistently vary as a function of children’s socio-economic status nor their school entry skills. That is, school uniforms did not address issues of economic and educational equality that have been at the center of much of the pro-uniform debates and the very reason many school officials and school systems require students to wear uniforms ( DeMitchell, 2006 ). Just as importantly, even with the large differences in the rates of school uniform mandates between public and private schools ( National Center for Educational Statistics, Digest of Education Statistics, 2013 ), we documented no differences in the outcomes of school uniform policies and student outcomes across different school sectors. Put another way, the associations between school uniforms and students’ socio-emotional, behavioral, and engagement outcomes were comparable (and in most instances close to zero) in both public and private schools. Taken together, what the results suggest is that the outcomes of school uniforms are far more similar than different for students of different backgrounds and for students enrolled in different types of schools across the United States.
With that said, one of the only consistent patterns that did emerge (and the third and final key theme) was that low-income children demonstrated fewer absences between first and fifth grade in schools with uniforms as compared with low-income children in schools without uniforms. The above is noteworthy given that absenteeism is at its highest point in the early elementary school years ( Ansari & Pianta, 2019 ) and there is long-standing evidence to suggest that low-income children are doubly at risk: They are more likely to be absent from school (Morrissey et al., 2013) and they are more likely to experience reduced learning due to absences as compared with their more advantaged peers ( Gershenson et al., 2017 ). Given the above, there has been longstanding interest in identifying in which contexts school absences are lowest, particularly for groups of vulnerable children (e.g., Rogers & Feller, 2018 ; Robinson et al., 2018 ). Accordingly, what the results of the present study suggest, is that school uniforms may be one context in which low-income students have fewer instances of absences; with that said, the mechanism for reduced absenteeism was not feelings of school belonging.
Despite these contributions to the extant literature, the results of the present investigation should be interpreted in light of some limitations. First and foremost, students were not randomly assigned to attend schools with and without uniforms. Consequently, our findings should be interpreted with caution as there might be unobservable confounds. With that said, a correlation is necessary for a causal effect and what our findings underscore is that there is no correlational support, on average, for school uniform policies; and in a few instances, the associations that did emerge between school uniform policies and student outcomes were in the opposite direction (i.e., negative). Second, although the present study presents a national snapshot of the outcomes of school uniforms between kindergarten and fifth grade, what is missing is an examination of longer-term outcomes in secondary school and beyond. In other words, our study cannot determine the potential associations for student outcomes of interest for middle or high school students. Additional research is needed to better understand the associations between school uniform policies and student outcomes in the older grades and. Third, because administrators in the ECLS-K were not asked about school uniforms on a yearly basis, the present study could not consider within school changes in uniform policies. Likewise, very few students changed between schools with and without uniforms and, consequently, we also could not examine within child change (i.e., child fixed effects models). For the above reasons, we limited or sample to students who remained in the same school between kindergarten and fifth grade to reduce any bias that may stem from school mobility or other unknown confounds. The above is of note because it limits the generalizability of our findings. Finally, our study relied largely on teacher reports of children’s socio-emotional skills because information from other sources was not consistently available across the elementary grades. The above is of note because our benchmarks for children’s socio-emotional development are based on teachers’ perceptions and may be biased. With that said, it is important to keep in mind that: (a) there are few, if any, large-scale and nationally representative studies that have more objective measures of socio-emotional development and (b) as part of our supplemental analyses, we also examined students’ own reports of their ties to their schools.
With these limitations and future directions in mind, the results of the present investigation bring a contemporary and national perspective to the ongoing debates surrounding school uniforms. In the main, despite the argument that school uniforms have the potential to improve students’ social-behavioral and engagement outcomes ( Daugherty, 2002 ; Kaiser, 1985 ; Stanley, 1996 ; Zernike, 2002 ), and shifting the process of reciprocal determinism (Grusec, 1994), the findings from the present study suggest otherwise: There is no conclusive evidence to suggest that students in schools with uniform mandates demonstrated stronger social-behavioral and engagement outcomes than students in schools without such mandates, and there was little evidence of heterogeneity. In fact, when examining students’ own reports of their school experiences, those in schools that required uniforms demonstrated higher levels of victimization and lower-levels of school belonging. If replicated with different samples and methods in the future, what these results suggest is that school uniforms may not be the most effective way to improve students’ social, behavioral, and engagement outcomes.
The authors acknowledge the support of the National Institute of Child Health and Human Development (R03 HD098420-02). Opinions reflect those of the authors and do not necessarily reflect those of the granting agency.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
IMAGES
VIDEO
COMMENTS
School uniforms: Do they really improve student achievement, behavior? This updated collection of research looks at how mandatory school uniforms impact student achievement, attendance and behavior as well as the presence of gangs in public schools.
Researchers are divided over how much of an impact — if any — dress policies have on student learning. There are multiple studies with conflicting conclusions, plus books such as 2018’s The Debate About School Uniforms, but the argument wears on, with a list of pros and cons on each side.
It explores three questions: What is the evidence for the impact of school uniform on students’ academic and health outcomes; what social, cultural and political rationales are made for uniform use; and what human rights may be affected by school uniform choice?
An examination of individual survey items revealed older students were more likely to report that school uniforms help to reduce bullying and teasing, and some students of high socioeconomic status reported that uniforms help reduce arguments with parents about clothing.
In general, students in schools that required school uniforms did not demonstrate better social skills, internalizing and externalizing behavior, or school attendance as compared with students in schools without school uniforms. These associations were true across both public and private schools.
Proponents have argued that school uniforms are a good thing for morale and community spirit. Some people have also claimed that uniform policies cause improvements in school attendance and academic achievement. And it’s easy to appreciate the reasoning.