How+do+uniforms+promote+professionalism+and+academic+acchievementclaire

= RAHS Note Page =

Source
http://web.ebscohost.com/ehost/detail?vid=11&hid=104&sid=bddacfe5-5601-4f6b-8dd5-ac087013382e%40sessionmgr104&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d%3d#db=keh&AN=12103384

Summary/Direct Quote
Uniforms and Academic Achievement

In their JER article, Brunsma and Rockquemore (1998) considered correlations between school uniforms and four variables for which data are available in NELS:88. Three variables were based on survey results of student recollection and self-report: (a) attendance (number of absences self-recollected); (b) behavior (composite of self-recollections of frequency of getting into trouble, of suspensions, and of physical fights); and (c) substance use (composite of six self-recollections dealing with frequency and amount of cigarettes, alcohol, and marijuana used over varying time periods). The fourth variable (d) academic achievement was used as a composite of a reading and a mathematics standardized test score, on the basis of school records. In this article, I focused on Brunsma and Rockquemore's claim that relates school uniform use to academic achievement. The authors presented their work as a contribution to discourse on uniforms in public schools:

Mandatory uniform policies have been the focus of recent discourse on public school reform. ... Tenth grade data from The National Educational [sic] Longitudinal Study of 1988 was used to test empirically the claims made by uniform advocates. ... (T)he authors found a negative effect of uniforms on student academic achievement. (Brunsma & Rockquemore, 1998, p. 53 Abstract)

NELS:88 classifies data by school sector: all schools, public schools, Catholic schools, and three types of private-but-not-Catholic schools, as shown in Table 1, extracted from Brunsma and Rockquemore's Table 1, and modified by inclusion of weighted distribution for omitted category--public schools. On the basis of a correlation matrix (Brunsma & Rockquemore, 1998, Appendix C), the authors stated, "Student uniforms are correlated slightly (.05) with standardized achievement scores, indicating a possible relationship" (p. 56). Next, they presented test score means: 52.89 for uniformed versus 50.58 for nonuniformed students and, using a t test, showed the 2.31-point difference significant at .01.

Thus, Brunsma and Rockquemore (1998) found that uniformed students have significantly higher test scores than do nonuniformed students. How, then, did they arrive at their claim of "negative effect of uniforms on student academic achievement?" The answer lies in misleading use of sector analysis. It is so unusual to find public school policy argued from data on a school sector that shows the opposite tendency, that it is worth examining the structure of their argument in detail.

Table 2 shows test score means for each school sector, extracted from Brunsma and Rockquemore's Table 2, modified by inclusion of an omitted category--public schools. Paradoxically (because the topic of their article is debate over uniforms in public schools), they did not mention test score means for the public school sector. However, inspection of data presented in Table 2, last column, indicates that uniformed public school students must have scored considerably higher than nonuniformed public school students in order for the differences between means of uniformed versus nonuniformed students in the various sectors (Catholic = -3.02 points, private-but-not-Catholic = +0.59 points, public = X) to produce the 2.31-point mean difference by which uniformed students outperformed nonuniformed students in the total sample. Although uniforms have been adopted by successful as well as failing public schools, they are more prevalent in urban schools (National Association of Elementary School Principals, 1998/2000), where they often are proposed as part of a comprehensive package of changes to improve failing schools. Therefore, it is unexpected that one finds positive correlation in the public sector between uniforms and test scores. Failure to even mention positive correlation in the public sector was the first step in the authors' transformation of positive correlation into their claim that uniforms are correlated negatively with achievement.

The second step in this transformation began with an acknowledgement of the positive correlation between uniforms and test scores for the total sample, but elided immediately into the following misleading use of sector analysis:

At the 10th-grade level, students wearing uniforms had significantly higher achievement (p < .01) than did students not wearing uniforms. That difference mirrored the hypothesized character of the difference as stated in the public discourse. However, when one breaks down this type of analysis into sectors, the relationships are not supported.

Catholic schools imply uniforms in most people's minds and in fact, Catholic schools account for 65.4% of all uniform policies--more than any other type of organization. (Brunsma & Rockquemore, 1998, p. 57)

The authors then focused exclusively on negative correlation between uniforms and test scores in the Catholic sector, never again mentioning the positive correlations in all other sectors and in the total sample. That Catholic schools "imply uniforms in most people's minds" provides no logical justification for substituting its atypical correlation for the positive correlation found in the total sample, yet that is what the authors did. Substitution of the negative correlation found in the single atypical sector for the positive correlation discovered in all other sectors and in the total sample was the second step in the authors' transformation of actual positive correlation into their claim that uniforms are correlated negatively with achievement.

If the direction and magnitude of correlation between uniforms and test scores for Catholic and public school sectors were similar, indicating the same processes might be operating in both, it could be appropriate to use the Catholic sector as an exploratory model for effects of uniforms in the public sector, particularly because the number of public school 10th graders in NELS:88 who wore uniforms might have been too small to achieve statistical significance, even with large correlation. When the correlation was in the opposite direction in the Catholic and public sectors, as in NELS:88, obviously the Catholic sector could not be used as a model for the public sector.

The authors made no attempt to explore possible contributions to the atypical correlation in the Catholic sector, a correlation that may have been accounted for by systematic regional effect on both variables (Bodine, 2003b). Instead, they concluded their consideration of uniforms and academic achievement by stating, "Most striking were the significant negative effects of uniforms on achievement, an outcome of much concern to educators and policy makers" (Brunsma & Rockquemore, 1998, p. 58).

That finding was premature attribution of substantive significance to statistical significance, or of causation to correlation, and was also highly misleading in light of the authors' conclusion that "students wearing uniforms had significantly higher achievement (p < .01) than did students not wearing uniforms" (Brunsma & Rockquemore, 1998, p. 57).