Selected Nonmusic Predictors of Urban Students’ Decisions to Enroll and Persist in Middle and High School Music Ensemble Electives

Abstract

The purpose of this study was to fit theoretical models of prediction to urban students’ decisions to enroll and persist in music ensembles using academic achievement, socioeconomic status (SES), number of parents/guardians at home, mobility, ethnicity, and sex as explanatory variables. Through multinomial logistic regression, I built predictive models for initial enrollment (i.e., 6th grade) and retention (8th and 10th grades) in band, string, and choir electives. At each grade level, predictive models supported differentiation between band, string, and choir students from nonmusic students on most factors as well as differences between instrumental and choir students. Choir students were differentiated from instrumental students in terms of academic achievement, SES, family structure, and mobility. These factors revealed more congruence with the population of nonmusic students than instrumental students. Factors influencing initial enrollment in band, string, and choir remained relatively stable over retention models, with notable exceptions: SES became a weaker predictor of band enrollment in high school, whereas number of parents/guardians at home became more salient for this group. All music participation was predicted by academic achievement; however, this was evidenced only in reading test scores for choir participants, whereas math and reading achievement predicted enrollments in instrumental music electives.

Keywords

school music participation music electives academic achievement urban students recruitment and retention

Despite egalitarian efforts within the profession to render music elective offerings at the middle and high school levels accessible and viable options for all students, researchers have found that enrollment in these electives is often highly associated with student demography (Elpus, 2013, 2014; Elpus & Abril, 2011; Fitzpatrick, 2006; Kinney, 2008, 2010; Lorah, Sanders, & Morrison, 2014; Stewart, 1991). A relatively recent demographic profile of music ensemble students in the United States provided by Elpus and Abril (2011), for example, illustrated that students who enroll in high school band, choir, and strings electives differ from nonmusic students on ethnicity, sex, socioeconomic status (SES), native language, standardized test scores, GPA, and parent education level. These findings echoed those of Stewart (1991), who found that high school music students from the senior class of 1982 came from mostly advantaged backgrounds and were predominantly female and white. Concerning in these data is that minorities, students of lower SES, English language learners, and students whose parents/guardians earned a high school diploma or less are underrepresented in music ensemble electives. Because findings from these studies suggest inequities in music elective enrollment over the past 30 years, further investigation into demographic and social factors influencing initial enrollment and retention in music performance electives would assist the profession in ascertaining how these factors are associated with student participation.

Previous research into performing ensemble enrollment and retention has primarily focused on instrumental music courses (i.e., band and strings). Because these courses require the purchase or rental of an instrument to participate, some have speculated that demographic variables, such as SES, limit participation in these classes to only those students who can afford instruments (see e.g., Albert, 2006; Elpus, 2013; Elpus & Abril, 2011; Fitzpatrick, 2006; Kinney, 2008, 2010; Phillips, 2003). Empirical findings in this area have largely supported this conjecture. For instance, McCarthy (1980) found that students from a higher SES participated in instrumental coursework longer than those from a lower SES. Klinedinst (1991) determined SES to be a more salient factor in predicting student retention in instrumental programs than academic competency or music aptitude. Wolfe (1969) found fewer students from lower SES homes beginning instrumental music in school. More recently, Kinney (2010) determined that SES was a salient factor in predicting retention but not initial enrollment in urban middle school band programs. Here, family structure, defined as single- or two-parent/guardian homes, and academic achievement were the most salient variables predicting initial enrollment in band, whereas family structure, academic achievement, SES, and sex (i.e., female) predicted retention. In regard to family structure and SES, Kinney’s findings resembled those of Corenblum and Marshall (1998) as well as Phillips (2003) and underscore complex concomitant relationships between school music participation and sociodemographic factors.

Academic achievement is also a factor associated with initial enrollment and persistence in instrumental music coursework. As noted previously, Kinney (2010) found academic achievement to be a significant predictor of both initial enrollment and retention in urban middle school band programs—students with higher academic achievement were more likely to enroll initially and persist. Likewise, other researchers have noted similar findings when comparing the academic achievement of music participants to nonparticipants (Catterall, 1997; Catterall. Chapleau, & Iwanaga, 1999; Elpus, 2013; Gouzouasis, Guhn, & Kishor, 2007; Miksza, 2007, 2010). Importantly, though, findings have suggested that differences between ensemble participants’ and nonparticipants’ academic achievement exist prior to enrollment decisions (Fitzpatrick, 2006; Kinney 2008) or are effectively mitigated when statistical controls for variables known to affect academic achievement, such as demography, prior academic achievement, unstructured time use (i.e., hours spent watching TV, playing video games, or working), and attitudes toward school, are employed (Elpus, 2013).

Although there have been several studies investigating factors influencing instrumental music participation in middle and high school, an investigation into the factors associated with initial string and choir ensemble enrollment and retention are rare in the literature. Ancillary findings of studies focusing on academic achievement and music elective participation have suggested that high school instrumentalists and chor-isters may be differentiated by sociodemographic factors (Elpus, 2013). Considering academic achievement as an example, it would be expected that choir students would score similarly to instrumental music students on measures of academic achievement if the sociodemographic factors known to impact test scores¹ were ubiquitous in both populations. This may not be the case. When disaggregating type of music participation, the higher academic achievement of ensemble participants evidenced in previous studies has been mostly limited to instrumental music students. Conversely, choir students’ academic achievement has been statistically equivalent to or lower than that of nonparticipants. Kinney (2008), for instance, found no significant differences between choir participants and students not participating in music and that both of these groups’ test scores were significantly lower than band participants both prior to and after music elective decisions had been made. Likewise, Elpus (2013) found that choir students scored significantly lower than nonmusic peers on the SAT. Moreover, he demonstrated that instrumental music students’ scores were upwardly biasing the bivariate relationship of music participation–SAT achievement in aggregate data.

Further complicating findings related to music student demography, much of the demographic data gathered about students participating in music electives do not differentiate among the various types of school music participation available to students. In large-scale studies investigating the demography of school music participants (e.g., Elpus & Abril, 2011; Lorah et al., 2014; Stewart, 1991), music elective participation has been defined dichotomously because of the nature of the data set employed.² As noted previously, findings from these studies have indicated differences between music and nonmusic students in regard to SES, family structure, ethnicity, and sex; however, the design of these studies does not allow for generalizations about specific music electives. Consequently, assuming a homogenous population of band, string, and choir students could bias the conclusion that all music students differ from nonmusic students. Further investigation using disaggregated definitions of music elective participation would help clarify and extend previous findings.

Because urban schools often enroll a greater proportion of minorities and economically disadvantaged students, music programs in urban schools are particularly at risk of failing to attract students to music elective offerings and being more likely to populate these offerings with students who differ from the general demographic composition of the school. Researchers have shown that music program enrollment in urban schools is affected by the perceived cultural relevance of the music program to the student (Albert, 2006; Doyle, 2014), parental involvement (Costa-Giomi & Chappell, 2007), family structure (Kinney, 2010), and ethnicity (Chenault, 1994). Mobility (i.e., student transience) has also been associated with retention in instrumental performing ensembles (Kinney, 2010). Thus, the association between student demography and music ensemble participation is particularly acute in urban schools.

Although researchers have focused on building a demographic profile of music students in general and instrumental music students specifically, little is known about the demography of students enrolled in other middle and high school music ensembles. Considering that choir participation does not typically involve the cost of instruments often associated with participating in an instrumental ensemble, it is possible that choir students differ in substantive ways from instrumentalists. It is also possible that continuous enrollment in choir (i.e., retention) could be affected by not having to purchase or rent an instrument to initiate or continue enrollment. To date, no research could be located that has examined this. Moreover, few researchers have investigated sociodemographic factors related to recruiting and retention in urban school programs, where nonmusic factors often mediate student participation in music. Given the lack of research with this population and the gaps identified in the literature, I undertook an investigation of selected sociodemographic factors that might be associated with students’ initial enrollment and retention in one urban school district’s middle and high school band, string, and choir music ensembles. Sociodemographic variables included in the study were selected based on their theoretical relevance to the research purpose and/or demonstrated significance in prior research. Thus, I sought to predict initial enrollment and retention in one urban school district’s middle and high school band, string, and choir elective classes using academic achievement, SES, family structure, mobility, ethnicity, and sex as predictor variables.

Method

Participants

The school district chosen for study was located in a midwestern metropolitan area. According to 2010 census data estimates, this area had a population of 787,033, with a median household income of $43,348. District enrollment totaled 49,602 and was comprised mainly of minority students (62.0% Black, 25.9% Caucasian, 8.9% Hispanic, 3.0% Asian, 0.2% Native American) across 116 schools. A majority of students were enrolled in free or reduced meal programs (78.5%), and 16.9% of students were considered transient. Females accounted for 51% of students. The district attendance rate was 91.7%, and the four-year graduation rate was 77%. The district comprised 16 middle schools (grades 6–8), 3 K–8 schools, 6 middle/high schools (6–12 and 7–12), and 18 high schools (grades 9–12). Although restricting the study to one district limited generalizability, the district was considered large enough to randomly distribute effects of recruiting/retention attributable to individual teachers.

Elective choices in this district began in 6th grade with students being able to choose general and fine art electives. Electives were referred to as “unified arts courses” and consisted of dance, drama, band, strings, choir, visual arts, health, and physical education. Full- and part-time licensed music teachers taught music elective courses, which met approximately five hours a week, on average. All electives from 6th through 12th grades were part of the school day and not “pull out” courses. Up to two electives could be taken during 8th grade for high school credit. One high school credit of fine arts, out of 24 total credits, was required to achieve a high school diploma of distinction.

A database containing student demographic information and achievement test scores was obtained through the assistance of a school district curriculum specialist after I received appropriate approvals from school district personnel and my university’s Institutional Review Board. I examined students’ music elective choices in 6th, 8th, and 10th grades for three independent cohorts, graduating in different years, to control for any potential cohort effects. Of the 6th-grade students (N = 12,104), 2,996 (24.8%) were enrolled in band, 820 (6.8%) in strings, and 2,164 (17.9%) in choir. Eighth grade (N = 11,679) consisted of 2,005 (17.2%) band, 518 (4.4%) string, and 1,917 (16.4%) choir students. For 10th grade (N = 13,581), 1,177 (8.7%) were in band, 495 (3.6%) in strings, and 1,244 (9.2%) in choir. Because the number of students participating in more than one ensemble was low (<1% for each grade level), these students were not retained in the data set for analysis.

Predictor Variables

Academic achievement

The database containing student demographic information provided reading and math achievement test raw scores for each participant. For this study, raw scores were considered the best measure of academic achievement to allow for comparisons among students with different academic tracking. The 6th- and 8th-grade achievement test was the state proficiency test. These assessments were administered annually in grades 3 through 8 to comply with No Child Left Behind policy. Assessments were designed specifically for Ohio students and based on the state’s academic content standards. For both reading and math, a score of 700 or above met the threshold for proficiency on a scaled score continuum of 555 to 851 for 6th-grade reading, 616 to 790 for 6th-grade math, 586 to 805 for 8th-grade reading, and 553 to 774 for 8th-grade math. The 10th-grade diagnostic measure was the state graduation test; it was the only standardized assessment for high school students. Developed by Ohio’s department of education, the graduation test was based on state content standards to measure students’ proficiency in core academic subjects. Scores 400 or above were considered proficient for both tests on a scaled score continuum of 262 to 552 for reading and 252 to 556 for math. The percentage of district students achieving proficiency for reading was 62.4% (6th grade), 71.1% (8th grade), and 78.0% (10th grade) and 45.6% (6th grade), 53.7% (8th grade), and 60.6% (10th grade) for math.

Socioeconomic status

Because the data set provided by the district did not include students’ annual household income, students’ free and reduced lunch (FRL) status was used as a relative indicator of SES. Although an imperfect measure, FRL has been an effective way to approximate students’ SES in previous studies when other information was not available (see Fitzpatrick, 2006; Kinney, 2008, 2010; Kinney & Forsythe, 2005; Nichols, 2003). Students qualifying for FRL were categorized as “lower SES” (6th graders, 80.3%; 8th graders, 75.0%; 10th graders, 69.8%), whereas those not qualifying were labeled as “higher SES” (6th graders, 19.7%; 8th graders, 25.0%; 10th graders, 30.2%).³

Family structure

Family structure was defined as the number of parents/guardians the school district had on record as living at a student’s primary residence. Several music studies have used this definition as an adequate indicator when investigating the correlation between academic achievement and family structure (Elpus, 2013; Elpus & Abril, 2011; Kinney, 2010). Students were categorized as those with single-parent/guardian family structures (6th graders, 53.6%; 8th graders, 56.3%; 10th graders, 53.1%) and those with two-parent/guardian family structures (6th graders, 46.4%; 8th graders, 43.7%; 10th graders, 46.9%).

Mobility

Mobility was divided into district mobility and school mobility for greater specificity. District mobility was defined as students moving into the district in the year immediately preceding elective enrollment decisions. For example, 6th-grade students were defined as transient in regard to district mobility if they moved into the district in 5th grade; 8th-grade students were defined as transient if they moved into the district in 7th grade. School mobility was defined as students who remained in the school district but transferred among schools beyond those associated with traditional matriculation through grade levels (e.g., moving to a high school building between 8th and 9th grades). Using these definitions, students were classified as having no district mobility (6th grade, 71%; 8th grade, 67.5%; 10th grade, 71.1%) or district mobility (6th grade, 29%; 8th grade, 32.5%; 10th grade, 28.9%) and no school mobility (6th grade, 74.8%; 8th grade, 73.9%; 10th grade, 69.5%) or school mobility (6th grade, 25.2%; 8th grade, 26.1%; 10th grade, 30.5%). Correlations between district and school mobility were found to be low (6th grade rφ = .14, 8th grade rφ = .11, 10th grade rφ = .14), further justifying their inclusion as separate predictors in the model.

Ethnicity and sex

Students whose ethnicities were classified as “Native American” or “other” were excluded from analyses because of a lack of representation (each <1%). With these students removed, ethnicity of 6th graders was White (25.6%), Black (69.0%), Hispanic (3.6%), and Asian (1.8%). Males accounted for 48.1% of 6th graders. Ethnicity of 8th-grade students was White (25.2%), Black (70.3%), Hispanic (2.3%), and Asian (2.1%). Males comprised 48.5% of 8th-grade students. Ethnicity of 10th-grade students was White (26.1%), Black (69.3%), Hispanic (2.4%), and Asian (2.2%). Males accounted for 48.8% of 10th-grade students.

Data Analysis

To determine the most parsimonious predictive models for the data, I employed polytomous (i.e., multinomial) logistic regression techniques to build preliminary and final models for each grade level.⁴ Predictor variables (N = 8) in this study were both categorical (i.e., SES, number of parent/guardians at home, district mobility, school mobility, ethnicity, and sex) and continuous (i.e., 6th-, 8th-, and 10th-grade reading and math achievement test scores). All categorical variables were dichotomous, with the exception of ethnicity, which was defined as White, Black, Hispanic, or Asian. Dummy (i.e., reference) coding for categorical variables was as follows: SES = higher SES, family structure = students from two-parent/guardian homes, district mobility = no district mobility, school mobility = no school mobility, ethnicity = White, and sex = male.

Analyses of data from the 6th-, 8th-, and 10th-grade participants employed a baseline-category logit model with the polytomous variable of music ensemble membership (i.e., band, strings, choir, or none) as the criterion variable, where the category, none, served as the baseline for comparison. Sample sizes of 12,104 (6th grade), 11,493 (8th grade), and 13,581 (10th grade) well exceeded Hosmer, Lemeshow, and Sturdivant’s (2013) sample size guidelines of a minimum number of 10 cases for each independent variable. Further, I determined that a sample size of 439 would be necessary to achieve a minimum power of .90 while keeping an alpha level of .05 for the test of each model. Thus, sample sizes for all cohorts were considered sufficient for ensuing analyses.

Results

Sixth-Grade Final Model

Predictor variables retained in the sixth-grade final model included: reading and math achievement test scores, SES, district mobility, school mobility, ethnicity, and sex. A test of this model versus a model with intercept only was statistically significant, χ²(27, N = 12,104) = 663.85, p < .001. To further examine the accuracy of the final model, El-Habil (2012) recommends comparing by-chance accuracy rates to the actual classification accuracy rates produced by the model. A 25% increase in predictive ability is used as a benchmark of a useful multinomial logistic regression model. For the sixth-grade final model, the overall classification rate was 52.7%, which exceeded a chance accuracy rate by 42.6%, thus satisfying the criterion. Table 1 shows regression coefficients, standard errors, Wald χ² test and associated probabilities, odds ratios, and 95% confidence intervals for the sixth-grade final model.

Table 1.

Final Model for Sixth-Grade Students.

Predictor	B	SE	Wald χ²	p	Odds Ratio	95% CI
Band
Reading	0.002	.001	2.14	.144	1.002	0.999, 1.005
Math	0.011	.001	76.98	<.001	1.011	1.008, 1.013
SES	0.348	.082	17.99	<.001	1.417	1.206, 1.664
District mobility	0.144	.071	4.16	.035	1.155	1.006, 1.327
School mobility	0.160	.068	5.54	.019	1.173	1.027, 1.340
Ethnicity^a
Black	−0.099	.075	1.72	.189	0.906	0.781, 1.050
Hispanic	−0.655	.237	7.66	.006	0.520	0.327, 0.826
Asian	−0.187	.176	1.13	.287	0.829	0.587, 1.171
Sex	−0.169	.063	7.29	.007	0.845	0.747, 0.955
Strings
Reading	0.009	.004	5.67	.017	1.009	1.002, 1.016
Math	0.010	.002	12.12	.001	1.011	1.008, 1.018
SES	0.314	.184	2.90	.089	1.368	0.954, 1.964
District mobility	0.501	.182	7.54	.006	1.650	1.154, 2.358
School mobility	0.265	.154	2.95	.086	1.304	0.963, 1.765
Ethnicity^a
Black	−0.386	.538	0.52	.473	0.680	0.237, 1.949
Hispanic	−1.630	1.022	2.55	.111	0.196	0.026, 1.451
Asian	0.400	.187	4.58	.032	1.492	1.034, 2.153
Sex	−0.609	.148	16.84	<.001	0.544	0.406, 0.727
Choir
Reading	0.006	.002	10.91	.001	1.006	1.002, 1.009
Math	−0.002	.001	1.79	.181	0.998	0.996, 1.001
SES	−0.033	.104	0.10	.753	0.968	0.790, 1.186
District mobility	0.064	.080	0.63	.426	1.066	0.911, 1.248
School mobility	0.039	.080	0.24	.626	1.040	0.889, 1.217
Ethnicity^a
Black	−0.551	.084	42.56	<.001	0.576	0.488, 0.680
Hispanic	−0.882	.292	9.10	.003	0.414	0.234, 0.734
Asian	−0.406	.200	4.13	.042	0.666	0.450, 0.986
Sex	−1.110	.077	207.00	<.001	0.329	0.283, 0.383

Note. Baseline comparison group is no music. CI = confidence interval; SES = socioeconomic status.

Comparison group is White.

Employing a .05 alpha level, several significant main effects surfaced. For band, significant main effects were associated with math scores, SES, district and school transience, ethnicity, and sex. Holding all other variables constant, each 10-point increase in math scores increased the odds of enrollment in band by 11%. Odds ratios (OR) and probabilities (P)⁵ for categorical explanatory variables indicated that while holding all other variables constant: (a) higher SES students were 42% more likely to enroll (P = .586); (b) Hispanic students were 48% less likely to enroll (P = .342), whereas no significant differences occurred for black or Asian students; (c) students with no district transience were 16% more likely to enroll (P = .536); (d) students with no school transience were 17% more likely to enroll (P = .540); and (e) males were 15.5% less likely to enroll (P = .458). Notably, CIs for all variables were relatively close to 1.00, with the exception of SES and Hispanic students, 95% CIs [1.206, 1.664] and [.327, .826], respectively. Reading test scores were not a significant predictor of initial band enrollment.

Reading scores, math scores, district mobility, ethnicity, and sex were significant predictors of string enrollment for sixth-grade students. Holding all other variables constant: (a) Each 10-point increase in math and reading scores increased the odds of enrollment in strings by 11% and 9%, respectively; (b) students with no district mobility were 65% more likely to enroll (P = .623); (c) males were 45.6% less likely to enroll (P = .352); and (d) Asian students were 49% more likely to enroll (P = .599). Reading and math CIs fell close to 1.00; however, the CIs of significant categorical predictors were substantially further from 1.00. SES and school mobility were not significant predictors of initial string enrollment.

Only reading test scores, sex, and ethnicity were significant predictors of initial choir enrollment. Each 10-point increase in reading scores increased the likelihood of choir enrollment by 6% when holding all other variables constant. Taking the reciprocal of the odds ratio for male choir enrollment (OR = .329) revealed that females were 3.04 times more likely to enroll in choir than males (P = .752). All minorities were significantly less likely to enroll in choir. Although CIs for reading scores and Asian students fell close to 1.00, all other significant predictors’ CIs deviated substantially further from 1.00. Math test scores, SES, district mobility, and school mobility were not significant predictors of initial choir enrollment.

Eighth-Grade Final Model

Predictor variables retained in the eighth-grade final model included: reading and math achievement test scores, SES, district mobility, school mobility, ethnicity, and sex. A test of this model versus a model with intercept only was statistically significant, χ²(27, N = 11,679) = 905.98, p < .001. The overall classification rate for the eighth-grade final model was 65.4%, which exceeded a chance accuracy rate by 35.3%, thus satisfying El-Habil’s (2012) criterion for model accuracy. Table 2 shows regression coefficients, standard errors, Wald χ² test and associated probabilities, odds ratios, and 95% confidence intervals for the eighth-grade final model.

Table 2.

Final Model for Eighth-Grade Students.

Predictor	B	SE	Wald χ²	p	Odds Ratio	95% CI
Band
Reading	.008	.001	42.29	<.001	1.009	1.006, 1.011
Math	.010	.001	45.31	<.001	1.010	1.007, 1.012
SES	.211	.061	12.01	.001	1.235	1.096, 1.391
District mobility	.282	.058	23.70	<.001	1.326	1.183, 1.485
School mobility	.256	.055	21.69	<.001	1.292	1.160, 1.439
Ethnicity^a
Black	−.031	.061	0.25	.614	0.969	0.860, 1.094
Hispanic	−.312	.177	3.09	.079	0.732	0.517, 1.036
Asian	−.575	.199	8.34	.004	0.563	0.381, 0.831
Sex	−.053	.052	1.04	.309	1.055	0.952, 1.168
Strings
Reading	.009	.004	4.97	.037	1.009	1.002, 1.018
Math	.013	.005	6.47	.011	1.013	1.003, 1.023
SES	.249	.232	1.15	.283	1.283	0.820, 1.024
District mobility	.023	.201	0.01	.910	1.023	0.689, 1.518
School mobility	.236	.196	1.45	.229	1.266	0.862, 1.859
Ethnicity^a
Black	−1.136	1.022	1.24	.267	0.321	0.043, 2.382
Hispanic	−.099	.614	0.03	.872	0.906	0.272, 3.020
Asian	.171	.227	0.57	.451	1.187	0.761, 1.851
Sex	−.957	.206	21.50	<.001	0.384	0.256, 0.575
Choir
Reading	.004	.001	10.11	.001	1.004	1.006, 1.011
Math	−.001	.001	0.90	.343	0.999	0.997, 1.001
SES	−.002	.066	0.00	.976	0.998	0.877, 1.136
District mobility	.020	.056	0.13	.722	1.020	0.915, 1.137
School mobility	−.032	.057	0.33	.569	0.968	0.867, 1.082
Ethnicity^a
Black	−.194	.062	9.69	.002	0.824	0.729, 0.931
Hispanic	−.372	.198	3.55	.059	0.689	0.468, 1.015
Asian	−.768	.210	13.38	<.001	0.464	0.307, 0.700
Sex	−.922	.055	276.24	<.001	0.398	0.357, 0.443

Note. Baseline comparison group is no music. CI = confidence interval; SES = socioeconomic status.

Comparison group is White.

Similar to the sixth-grade model, math scores, SES, district and school transience, and ethnicity were significant predictors of band enrollment; however, in the eighth-grade model, reading test scores were also significant. Examining odds ratios for these variables revealed that (a) each 10-point increase in math and reading test scores increased the odds of band enrollment by 10% and 9%, respectively; (b) students from a higher SES were 24% more likely to be enrolled (P = .553); (c) students with no district mobility were 33% more likely to be enrolled (P = .570); and (d) students with no school mobility within the district were 29% more likely to be enrolled (P = .564). Unlike the sixth-grade model where only Hispanic students were significantly less likely to enroll, the only significant predictor within ethnicity for the eighth-grade model was Asian students, with 44% less likely to enroll in band (P = .360). Also contrary to the sixth-grade model, there was no significant effect of sex on band enrollment. CIs for significant predictors were relatively small and close to 1.00, with the exception of Asian students, 95% CI [.381, .831].

Sex and reading and math test scores were significant predictors of eighth-grade enrollment in strings. Females were 2.60 times more likely to enroll (P = .722), and each 10-point increase in reading and math test scores increased odds of string enrollment by 9% and 13%, respectively. Although CIs for reading and math were close to 1.00, the CI for sex fell further away, 95% CI [.256, .575]. No other explanatory variable proved to be a significant predictor of string enrollment at the eighth-grade level.

Like the sixth-grade model, reading test scores, ethnicity, and sex were significant predictors of choir enrollment for eighth-grade students. Females were 2.51 times more likely to enroll (P = .715), and black and Asian students were significantly less likely to enroll. Hispanic students were, again, less likely to enroll in choir, although this finding fell just above the criterion for statistical significance. As with the sixth-grade model, reading test scores in eighth grade were a predictor of choir enrollment, with each 10-point increase in test scores increasing the odds of enrollment by 4%, although the CI for this variable was close to 1.00. Math scores, SES, district mobility, and school mobility were not significant predictors of choir enrollment for eighth graders.

10th-Grade Final Model

The 10th-grade final multinomial logistic regression model utilized all predictor variables because none were eliminated through preliminary analyses. A test of this model versus a model with intercept only was statistically significant, χ²(30, N = 13,581) = 1061.25, p < .001. The overall classification rate for the final model was 85.5%, which exceeded a chance accuracy rate by 25.7%, thus satisfying El-Habil’s (2012) criterion for model accuracy. Table 3 shows regression coefficients, standard errors, Wald χ² test and associated probabilities, odds ratios, and 95% confidence intervals for the 10th-grade final model.

Table 3.

Final Model for 10th-Grade Students.

Predictor	B	SE	Wald χ²	p	Odds Ratio	95% CI
Band
Reading	.005	.001	17.06	<.001	1.005	1.003, 1.008
Math	.010	.001	103.33	<.001	1.010	1.008, 1.012
SES	.038	.049	0.61	.433	1.039	0.944, 1.142
Family structure	.190	.046	16.79	<.001	1.209	1.104, 1.324
District mobility	.138	.051	7.30	.007	1.148	1.039, 1.269
School mobility	.083	.051	2.68	.101	1.087	0.984, 1.200
Ethnicity^a
Black	−.221	.051	18.91	<.001	0.801	0.725, 0.886
Hispanic	−.362	.150	5.87	.015	0.696	0.519, 0.933
Asian	−.100	.140	0.51	.477	0.905	0.688, 1.191
Sex	.026	.045	0.33	.567	1.026	0.940, 1.120
Strings
Reading	.002	.004	0.17	.683	1.002	0.994, 1.010
Math	.008	.003	5.00	.025	1.008	1.001, 1.014
SES	.128	.158	0.65	.420	1.136	0.833, 1.550
Family structure	.085	.152	0.31	.576	1.089	0.808, 1.466
District mobility	.145	.159	0.83	.363	1.156	0.846, 1.577
School mobility	.465	.175	7.07	.008	1.592	1.130, 2.242
Ethnicity^a
Black	−.348	.165	4.41	.036	0.706	0.511, 0.977
Hispanic	−.037	.431	0.01	.931	0.963	0.414, 1.543
Asian	.075	.401	0.04	.852	1.077	0.491, 2.362
Sex	−.722	.155	21.72	<.001	0.486	0.359, 0.658
Choir
Reading	.003	.001	4.57	.032	1.003	1.000, 1.005
Math	−.002	.001	5.44	.020	0.998	0.995, 1.000
SES	−.100	.048	4.38	.036	0.905	0.825, 0.994
Family structure	.074	.045	2.75	.097	1.077	0.987, 1.176
District mobility	.019	.048	0.16	.688	1.019	0.928, 1.119
School mobility	.027	.048	0.31	.576	1.027	0.935, 1.129
Ethnicity^a
Black	−.045	.052	0.76	.384	0.956	0.863, 1.058
Hispanic	−.526	.183	8.24	.004	0.591	0.413, 0.846
Asian	−.309	.161	3.69	.055	0.734	0.536, 1.006
Sex	−.903	.047	373.20	<.001	0.406	0.370, 0.444

Note. Baseline comparison group is no music. CI = confidence interval; SES = socioeconomic status.

Comparison group is White.

Similar to the eighth-grade model, math and reading test scores were significant predictors of band enrollment with each 10-point increase in math and reading scores increasing odds of enrollment by 10% and 5%, respectively. District mobility was also a significant predictor—those with no district mobility were 15% more likely to enroll (P = .534). SES, which had been a significant predictor of band enrollment in 6th and 8th grades, was not a significant predictor for the 10th-grade model (p = .433). Instead, family structure emerged as a significant predictor of band enrollment. Those from two-parent/guardian families were 21% more likely to be enrolled (P = .547). Also noteworthy, black and Hispanic students were significantly less likely to enroll in band programs. CIs for all predictor variables were relatively close to 1.00. Neither school mobility nor students’ sex was a significant explanatory factor for band enrollment in the 10th-grade final model.

Again, sex was a significant predictor for enrollment in strings, with females being 2.06 times more likely to enroll (P = .673). As in 6th- and 8th-grade final models, math test scores were a significant predictor of string enrollment, with each 10-point increase in test scores increasing the odds of enrollment by 8%. School mobility and ethnicity also achieved a level of significance in predicting string enrollment. Students not transferring among schools in the district were 59% more likely to be in string ensembles (P = .614), and black students were 29.4% less likely to enroll (P = .414). With the exception of sex and school mobility, 95% CIs [0.359, 0.658] and [1.130, 2.242], respectively, CIs for other significant predictors were close to 1.00. Reading test scores, SES, family structure, and district mobility did not contribute significantly to predictions of string enrollment in 10th grade.

Explanatory variables significantly predicting enrollment in 10th-grade choir consisted of reading and math test scores, SES, ethnicity, and sex. Findings for reading test scores and sex were similar to 6th- and 8th-grade final models. Females were 2.46 times more likely to be in choir (P = .711), and for every 10-point increase in reading scores, odds of choir membership increased by 3%. Unique to the 10th-grade final model, math scores and SES were significant predictors of choir membership. Importantly, though, these were inverse relationships. As math scores increased by 10 points, the odds of enrolling in choir decreased by 2%. Students from a higher SES were 9.5% less likely to be in choir (P = .475). For minorities, only Hispanic students were significantly less likely to enroll, although it should be noted that Asian students fell just above the null rejection criterion. Unlike 6th- and 8th-grade final models, black students were almost as likely to be in choir as white students (OR = .956, p = .384). With the exception of sex, 95% CI [0.370, 0.444], CIs for predictors were close to 1.00, with two being bounded by 1.00 on either the low or high extreme, namely, reading 95% CI [1.000, 1.005] and math 95% CI [0.995, 1.000]. Family structure, district mobility, and school mobility were not significant predictors of choir enrollment at the 10th-grade level.

Comparisons Among Band, String, and Choir Students

Multinomial logistic regression also allows for comparisons among all categories of a polytomous dependent variable by changing the category of reference in the baseline logit model. In an effort to draw conclusions about how band, string, and choir enrollments in 6th, 8th, and 10th grades might differ, I used a series of multinomial regressions, changing the reference category so that all possible baseline comparisons could be made without redundancy. These models were identical to the final models previously described in terms of predictor variable inclusion. Only findings achieving statistical significance are discussed in the following.

For sixth grade, comparing string to band enrollment revealed significant differences only for ethnicity and sex. Males were 55% (P = .608) more likely to enroll initially in band than strings, Wald χ² = 8.44, p = .004. Asian students were 64% (P = .621) more likely to enroll initially in strings than band, Wald χ² = 6.97, p < .008. Examining predictors for string and choir enrollment showed that females were 65% (P = .623) more likely to enroll initially in choir than strings, Wald χ² = 9.92, p < .002, and black students were 61% (P = .617) less likely to enroll initially in choir than strings. Band and choir initial enrollment differed significantly with regard to math test scores, SES, ethnicity, and sex. Holding other variables constant, odds of enrolling in band, as compared to choir (a) increased by 2% for every 10-point increase in math scores, Wald χ² = 65.07, p < .001; (b) were 46% (P = .593) greater for students from a higher SES, Wald χ² = 12.60, p < .001; (c) were 57% (P = .611) greater for black students, Wald χ² = 24.14, p < .001; and (d) were 2.56 (P = .719) times greater for males, Wald χ² = 124.32, p < .001.

For eighth grade, string and band enrollment was differentiated only by sex. Here, males were 2.75 (P = .733) times more likely to enroll in band than strings, Waldχ² = 23.16, p < .001. Only math test scores differentiated string and choir enrollment, where the odds of enrolling in strings increased by 14% for every 10-point increase in math scores (p < .001). Reading, math, SES, district mobility, school mobility, and sex were significant in predicting band versus choir enrollment: (a) The odds of enrolling in band increased by 11% and 5% for every 10-point increase in math scores, Wald χ² = 40.63, p < .001, and reading scores, Wald χ² = 9.78, p = .002, respectively; (b) students from a higher SES were 24% (P = .554) more likely to enroll in band, Wald χ² = 7.21, p = .007; (c) those with no district or school transience were 30% (P = .565) and 33% (P = .571) more likely to enroll in band, Wald χ² = 13.18, p < .001 and Wald χ² = 16.84, p < .001, respectively; and (d) males were 2.65 (P = .726) times more likely to enroll in band than choir, Wald χ² = 202.04, p < .001.

In the 10th-grade final model, string and band enrollment was differentiated only by sex, with males being 2.11 (P = .678) times more likely to enroll in band, Wald χ² = 21.87, p < .001. Like the 8th-grade model, only math scores differentiated string and choir enrollment, with enrollment in strings increasing by 10% for every 10-point increase in math scores, Wald χ² = 8.19, p = .004. Only math scores and sex significantly differentiated band from choir enrollment in the 10th-grade model. Here, the odds of band enrollment increased by 13% for every 10-point increase in math scores, Wald χ² = 84.39, p < .001, and males were 2.53 (P = .717) times more likely to enroll in band, Wald χ² = 226.48, p < .001.

Discussion

The purpose of this study was to fit theoretical models of prediction to urban students’ decisions to enroll and persist in music ensemble electives using academic achievement, SES, family structure, district mobility, school mobility, ethnicity, and sex as explanatory variables. Through multinomial logistic regression techniques, I built three final models that best fit these data for initial enrollment (i.e., 6th grade) and retention (8th and 10th grade) in band, string, and choir ensemble electives. These models extend previous findings related to sociodemographic factors associated with music ensemble participation by allowing comparisons of band, string, and choir students to the general student population as well as each other across middle school and high school elective decisions.

Consistent with previous research (Elpus, 2013; Kinney, 2008, 2010), findings related to academic achievement generally support the conclusion that higher achieving students enroll initially and persist in instrumental music electives. Math achievement in particular was a significant predictor of initial enrollment and persistence into 8th- and 10th-grade band and string programs. That this variable proved consistent across the three grade levels examined in this study lends further support to previous findings suggesting that higher achieving students are attracted to instrumental programs from the outset and that systematic differences between this population and the general school population remain relatively stable over time (Correnblum & Marshall, 1998; Elpus, 2013; Fitzpatrick, 2006; Kinney, 2008, 2010). Interestingly, achievement in reading was significant in predicting 8th- and 10th-grade enrollment in band programs but not initial enrollment. A similar anomaly occurred for strings, where enrollment in 6th and 8th grades was predicted by reading scores; yet, reading test scores did not predict enrollment in 10th grade. It should be noted that while the regression coefficients for these two anomalies revealed a positive relationship, each failed to achieve a level of significance.

Results pertaining to choir students and academic achievement were mixed. In terms of reading achievement, choir students’ test scores were significantly higher than those of nonmusic students in 6th, 8th, and 10th grades. This stands in contrast to Kinney (2008), who found no significant differences between choir students and the general population in regard to reading achievement. Although these results do not necessarily support the significant gains in reading achievement based on choir participation reported by Hearnsberger (2013), such findings are encouraging and suggest that higher achieving students may be attracted to all music ensemble programs, not just instrumental offerings. In the case of choir, it could be that an affinity for language, as expressed through lyrics, draws higher achieving students in reading to the subject. Researchers seeking to extend our understanding of factors influencing enrollment decisions in choir are encouraged to consider this in future research endeavors.

No differentiation occurred between choir students and the general population for math achievement scores, except in 10th grade, where lower scores in math significantly predicted choir enrollment. These findings are more consistent with previous literature than the findings related to reading (Elpus, 2013; Kinney, 2008). Taken alone, this reinforces that the academic achievement of choir students is more similar to nonmusic students than their instrumental music counterparts; however, it is important to note that some studies have not disaggregated academic achievement into verbal and quantitative measures. Elpus (2013), for example, considered composite SAT scores, which were comprised of both reading and mathematics, and reported a 10-point deficit for choir students versus nonmusic students. Results of the present investigation suggest choir students’ composite test scores might be negatively skewed because of math performance and should be disaggregated in terms of subject matter in future studies.

Examining academic achievement further revealed that band and string students were differentiated from choir students on measures of math and reading achievement. In cases where significant differences occurred, math and reading scores were higher for those enrolled in band and strings. Again, these findings are consistent with previous research (Elpus, 2013; Kinney 2008) and further reinforce the notion that instrumental students’ test scores upwardly bias the relationship between music participation and academic achievement.

When considering the impact of SES on enrollment and retention, it appears that this predictor functioned somewhat differently for band and string students. For band students, those from a higher SES were more likely to enroll initially. Higher SES also significantly predicted 8th-grade band enrollment, yet it was not a factor in predicting 10th-grade band enrollment. These findings are somewhat contradictory to Kinney (2010), who found SES to predict persistence in band programs but not initial enrollment. As many have suggested (Albert, 2006; Elpus, 2013; Elpus & Abril, 2011; Fitzpatrick, 2006; Kinney, 2008, 2010; Lorah et al., 2014; Phillips, 2003), the associated cost of procuring an instrument to participate in band programs oftentimes precludes the participation of students from families with lower incomes. In these data, this seems to be the case, at least for initial enrollment in band and persistence into 8th grade. On the other hand, SES was not a factor for predicting string enrollments in any grade. If the cost of procuring an instrument is indeed a factor, this did not manifest for string students or high school band students. In the case of strings, other factors, such as ethnicity, school mobility, and sex, were significant predictors of 10th-grade enrollment in these data. Perhaps these factors serve to override SES for string students. Moreover, lower student enrollment in strings could have allowed for more school-owned instruments to be distributed to students in this study. Likewise, the high school band programs included in this investigation might have had enough school-owned instruments to assist low-income students in overcoming this financial barrier, especially considering that student participation in band diminished from 6th grade (n = 2,005) to 10th grade (n = 1,177). Researchers are encouraged to examine how instrumental directors teaching in economically depressed settings contend with the financial burden oftentimes associated with participation in their classes.

Choir students were unique in regard to SES. For each grade level, choir had a more equitable proportion of lower SES students than band and strings. Results for 10th grade in particular showed lower SES students significantly more likely to enroll in choir, though the odds ratio for this was relatively small (i.e., OR = 1.10, P = .475). Such differentiation between populations could also be responsible for the differences associated with academic achievement found in this study. Considering that SES is one of the most salient factors associated with academic achievement (Zwick & Grief Green, 2007), a larger proportion of lower SES students in choir could potentially explain finding no differences or significantly lower scores on math in this study. What is promising, though, is that reading scores did not follow this trend. Finding significantly higher reading scores at each grade level despite a greater proportion of lower SES students represented in choir is a unique and encouraging finding. Perhaps, as Johnson and Eason (2016) have noted, what is being evidenced by these data is increased school engagement because of participation in music, which then leads to higher academic achievement. Determining why this occurred only for reading scores and if participation in choir had any direct or ancillary effect on this outcome would be a welcome addition to the literature.

Family structure was significant only in the 10th-grade model for band students. Although significant findings related to family structure have not been replicated in studies utilizing national data sets (cf. Elpus & Abril, 2011), Kinney (2010) found that it was a more salient predictor for retention of urban band students than SES in later grades. Like Kinney’s findings, students in the present study from two-parent/guardian families were more likely to persist—in this case, into high school band programs. That family structure did not predict enrollment for band students at the middle school level or for string and choir students at any grade level could point to the rigorous schedule often associated with participation in high school band. Because high school band typically includes marching band as well as other before- and after-school activities (e.g., jazz band, pep band), what may be evidenced here is the difficulty single parents have in navigating these additional activities.

Both district and school mobility impacted enrollment in music electives as well. Interestingly, though, when these variables manifested significant effects, their influence was limited to instrumental ensembles only. With the exception of 10th-grade school mobility, band enrollment was contingent on district and school mobility in all other instances. Likewise, string enrollment showed marked tendencies toward students with little to no transience. What seems clear from these findings is that instrumental music programs recruit and retain students who typically remain in the district and do not transfer among schools. Choir students, on the other hand, were not differentiated from nonmusic students on this factor. In fact, in 8th grade, choir students were significantly more likely to be transient than band students. Given that transience is often linked with financial insecurity, single-parent/guardian family structures, parental support, and lower academic achievement in urban teaching environments (Ingersoll, Scamman, & Eckerling, 1989; Kerbow, 1996; Schuler, 1990), these findings provide a broader understanding of the differentiated nature between instrumental and choir students in urban school systems.

With the exception of a larger proportion of Asian students enrolled in string programs, regression coefficients for minority enrollment in band, string, and choir programs indicated negative relationships. In cases where significance was achieved, disproportionately fewer numbers of minorities enrolled. Somewhat troubling is that in certain cases, like band, initial enrollment was not as associated with ethnicity, whereas later enrollment was. Compared to other findings (Elpus & Abril, 2011; Stewart, 1991), these data also suggest that music ensemble electives are often not reflective of the general student population and may not be resonating with certain minority groups (see also Doyle, 2014). This is obviously a concern for the profession. Music teachers are encouraged to consider this when developing curricular offerings, especially in school systems where a greater proportion of minority students exist.

Students’ sex predicted initial enrollment in all music ensembles. Specifically, females were more likely to enroll than males in each group, with choir being the dominating choice. Notably, band was the only enrollment to become more balanced between males and females in 8th and 10th grades. Although previous research has noted disproportionate numbers of females in all music ensemble electives (Kinney, 2010; McCarthy, 1980; Zervoudakes & Tanur, 1994), some researchers have shown equivalent or male-dominated enrollments in band (Fortney, Boyle, & DeCarbo, 1993). As Kinney (2010) and Kessels (2005) have noted, enrollment in music ensembles may be contingent on gender-stereotyped perceptions adolescents have toward music as a serious course of study. Because self-identification and social roles of children are so important to adolescent development (see Cramer, Million, & Perreault, 2002; North & Hargreaves, 1999), it is imperative that music teachers take these issues into account when positioning their courses for the school population at large.

The present study was among the first to undertake a systematic comparison of nonmusic factors influencing initial enrollment and persistence in band, string, and choir ensemble electives. Factors influencing initial enrollment in these electives remained relatively stable over both retention models, with some notable exceptions. SES, for example, became less of a factor in influencing the persistence of band students into high school, whereas the number of parents/guardians at home became a significant factor for this group in 10th grade. District and school mobility became more salient for enrollment in string programs from initial enrollment to high school. SES also became a salient predictor of choir enrollment in 10th grade, although it did not predict enrollment in 6th or 8th grades. Notably, the results for choir showed that students from a lower SES were more likely to be enrolled at the 10th-grade level.

Findings from the study support the conclusion that instrumental students differ from choir students in terms of academic achievement, SES, family structure, and mobility. For choir, these factors suggested more congruence with the population of nonmusic participants than instrumental students. This finding has implications for both researchers and teachers. Researchers examining the concomitant relationship between music participation and academic achievement must consider the limitations of working with aggregate data. As others (Elpus, 2013; Fitzpatrick, 2006; Kinney, 2008, 2010) and now this study have shown, differences in academic achievement associated with music participation may be more of a product of demography and selection than an effect of music instruction. Researchers are also encouraged to consider the limitations associated with composite scores of academic achievement given the differentiation between choir and instrumental students on math and reading found in this study. For teachers, this finding could reflect enrollment biases inadvertently created by the curricular structure of ensembles. By design, most instrumental ensembles do not accommodate beginners past middle school, whereas choir does. Consequently, students seeking to fulfill high school arts requirements could be participating in choir because of limited alternatives. If choir is serving this function, it would follow that its demography is more reflective of the general student body. A longitudinal study of music ensemble enrollment and dropout trends might help clarify this.

Given the limited sample frame employed in the study, especially in regard to student ethnicity and the underrepresentation of Hispanic students in the data set, generalizations are cautioned. Readers are also advised to consider the attrition in music ensemble enrollment occurring from 6th to 10th grade in these data, which could have been partially responsible for creating a distillation among the variables most affecting enrollment decisions and rendered certain ensemble participants’ demography more ubiquitous. Further, although many predictors were found to significantly contribute to enrollment decisions, corresponding CIs for several predictors deviated relatively little away from 1.00 and should be interpreted accordingly. For example, odds ratio CIs for 10th-grade choir students were close to 1.00 for every significant predictor except sex, suggesting relative ubiquity among choir participants and nonmusic participants for these variables. Replication of these findings utilizing diverse data sets would go a long way in reinforcing these factors’ relative association with student enrollment in music electives.

With these caveats, music educators are encouraged to consider these models as codifying certain sociodemographic factors associated with initial enrollment decisions and persistence in music ensemble electives. Making a concerted effort to recruit and retain minority students, males, those from a lower SES, and transient students will go far in bolstering enrollment in music ensemble electives, especially if these efforts are coupled with strategies to make these experiences more relevant to traditionally underserved populations. The effect of the individual teacher on program success was not considered in this study but is obviously a factor in populating these ensembles. Teachers of each respective music elective offering are thus encouraged to consider how nonmusic factors function in the context of the models presented previously. Through deliberate, conscientious efforts to reach students often underserved by music ensemble offerings, teachers will no doubt create a more democratic, equitable, and viable elective choice for all.

Supplemental Material

DS_10.1177_0022429418809972 – Supplemental material for Selected Nonmusic Predictors of Urban Students’ Decisions to Enroll and Persist in Middle and High School Music Ensemble Electives

Supplemental material, DS_10.1177_0022429418809972 for Selected Nonmusic Predictors of Urban Students’ Decisions to Enroll and Persist in Middle and High School Music Ensemble Electives by Daryl W. Kinney in Journal of Research in Music Education

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.

Notes

Supplemental Material

Appendices A through E are available in the online version of the journal at

Author Biography

Daryl W. Kinney is associate professor of music education at The Ohio State University. His research interests include music and nonmusic corollaries of formalized music instruction, music perception/cognition, and music teacher education.

References

Albert

D. J.

(2006). Strategies for the recruitment and retention of band students in low socioeconomic school districts. Contributions to Music Education, 33, 53–72.

Card

Rothstein

(2007). Racial segregation and the Black-White test score gap. Journal of Public Economics, 91(11/12), 2158–2184. doi:10.1016/j.jpubeco.2007.03.006

Catterall

J. S.

(1997). Involvement in the arts and success in secondary school. Washington, DC: Americans for the Arts.

Catterall

Chapleau

Iwanaga

(1999). Involvement in the arts and human development: General involvement and intensive involvement in music and theatre arts. In Fiske

E. B.

(Ed.), Champions of change: The impact of the arts on learning (pp. 1–18). Washington, DC: Arts Education Partnership.

Chenault

J. B.

(1994). A survey of public school music in the state of North Carolina for 1990–1991. Dissertation Abstracts International, 45(07), 2022.

Corenblum

Marshall

(1998). The band played on: Predicting students’ intentions to continue studying music. Journal of Research in Music Education, 46, 128–140. doi:10.2307/3345765

Costa-Giomi

Chappell

(2007). Characteristics of band programs in a large urban school district: Diversity or inequality? Journal of Band Research, 42, 1–18.

Cramer

K. M.

Million

Perreault

L. A.

(2002). Perceptions of musicians: Gender stereotypes and social role theory. Psychology of Music, 30, 164–174. doi:10.1177/0255761407085646

Doyle

J. L.

(2014). Cultural relevance in urban music education: A synthesis of the literature. Update: Application of Research in Music Education, 32, 44–51. doi:10.1177/8755123314521037

10.

El-Habil

A. M.

(2012). An application on multinomial logistic regression model. Pakistan Journal of Statistics and Operation Research, 8, 271–291 doi:10.1234/pjsor.v8i2.234

11.

Elpus

(2013). Is it the music or is it selection bias? A nationwide analysis of music and nonmusic students’ SAT scores. Journal of Research in Music Education, 61, 175–194. doi:10.1177/0022429413485601

12.

Elpus

(2014). Evaluating the effect of no child left behind on U.S. music course enrollments. Journal of Research in Music Education, 62, 215–233. doi:10.1177/0022429414530759

13.

Elpus

Abril

C. R.

(2011). High school music ensemble students in the United States: A demographic profile. Journal of Research in Music Education, 59, 128–145. doi:10.1177/0022429411405207

14.

Fitzpatrick

K. R.

(2006). The effect of instrumental music participation and socioeconomic status on Ohio fourth-, sixth-, and ninth-grade proficiency test performance. Journal of Research in Music Education, 54, 73–84. doi:10.1177/002242940605400106

15.

Fortney

P. M.

Boyle

J. D.

DeCarbo

N. J.

(1993). A study of middle school band students’ instrument choices. Journal of Research in Music Education, 41, 28–39. doi:10.2307/3345477

16.

Freedle

R. O.

(2003). Correcting the SAT’s ethnic and social-class bias: A method for reestimating SAT scores. Harvard Educational Review, 73, 1–43.

17.

Gouzouasis

Guhn

Kishor

(2007). The predictive relationship between achievement and participation in music and achievement in core grade 12 academic subjects. Music Education Research, 9, 81–92. doi:10.1080/14613800601127569

18.

Hearnsberger

K. A.

(2013). The effect of choral music education on literacy achievement. Doctoral dissertation, University of Arkansas at Little Rock. Available from ProQuest Dissertations & Theses A&I; ProQuest Dissertations & Theses Global. (1427342671). Retrieved from http://nf4hr2ve4v.search.serialssolutions.com.proxy.lib.ohio-state.edu/docview/1427342671?accountid=9783

19.

Hedges

Nowell

(1995). Sex differences in mental test scores, variability, and numbers of high-scoring individuals. Science, 269(5220), 41–45. doi:10.1126/science.7604277

20.

Hosmer

D. W.

Lemeshow

Sturdivant

R. X.

(2013). Applied logistic regression (3rd ed.). Hoboken, NJ: John Wiley & Sons, Inc.

21.

Ingersoll

G. M.

Scamman

J. P.

Eckerling

W. D.

(1989). Geographic mobility and student achievement in an urban setting. Educational Evaluation and Policy Analysis, 11, 143–149. doi:10.3102/01623737011002143

22.

Johnson

C. M.

Eason

B. J. A.

(2016). Evaluation of the impact of music program participation on students’ musical and academic success, and school engagement in the metropolitan Nashville Public Schools: A comprehensive test of pathways and contextual factors. Retrieved from http://hdl.handle.net/1808/20615

23.

Kerbow

(1996). Patterns of urban school mobility and local school reform. Journal of Education for Students Placed at Risk, 1, 147–169. doi:10.1207/s15327671espr0102_5

24.

Kessels

(2005). Fitting into the stereotype: How gender-stereotyped perceptions of prototypic peers relate to liking for school subjects. European Journal of Psychology of Education, 20, 309–323. doi:10.1007/BF03173559

25.

Kinney

D. W.

(2008). Selected demographic variables, school music participation and achievement test scores of urban middle school students. Journal of Research in Music Education, 56, 145–161. doi:10.1177/0022429408322530

26.

Kinney

D. W.

(2010). Selected nonmusic predictors of urban students’ decisions to enroll and persist in middle school band programs. Journal of Research in Music Education, 57, 334–350. doi:10.1177/0022429409350086

27.

Kinney

D. W.

Forsythe

J. L.

(2005). The effects of Arts IMPACT school curriculum upon the Ohio fourth-grade proficiency test scores. Bulletin of the Council for Research in Music Education, 164, 35–48.

28.

Klinedinst

(1991). Predicting performance achievement and retention of fifth-grade instrumental students. Journal of Research in Music Education, 39, 225–238. doi:10.2307/3344722

29.

Lorah

J. A.

Sanders

E. A.

Morrison

S. J.

(2014). The relationship between English language learner status and music ensemble participation. Journal of Research in Music Education, 62, 234–244. doi:10.1177/0022429414542301

30.

McCarthy

(1980). Individualized instruction, student achievement, and dropout in an urban elementary instrumental music program. Journal of Research in Music Education, 28, 59–69. doi:10.2307/3345053

31.

McLanahan

S. S.

(1985). Family structure and the reproduction of poverty. American Journal of Sociology, 90, 873–901. doi:10.2307/2779522

32.

Miksza

(2007). Music participation and socioeconomic status as correlates of change: A longitudinal analysis of academic achievement. Bulletin of the Council for Research in Music Education, 172, 41–58.

33.

Miksza

(2010). Investigating relationships between participation in high school music ensembles and extra-musical outcomes: An analysis of the Educational Longitudinal Study of 2002 using a bioecological development model. Bulletin of the Council for Research in Music Education, 186, 7–25.

34.

Nichols

(2003). Prediction indicators for student failing the state of Indiana high school graduation exam. Preventing School Failure, 47, 112–120. doi:10.1080/10459880309604439

35.

North

A. C.

Hargreaves

D. J.

(1999). Music and adolescent identity. Music Education Research, 1, 75–92. doi:10.1080/1461380990010107

36.

Phillips

(2003). Contributing factors to music attitude in sixth-, seventh-, and eighth-grade students. Unpublished doctoral dissertation, University of Iowa, Iowa City.

37.

Rumberger

R. W.

(2003). The causes and consequences of student mobility. The Journal of Negro Education, 72, 6–21.

38.

Schuler

D. B.

(1990). Effects of family mobility on student achievement. ERS Spectrum, 8, 17–24.

39.

Stewart

(1991). Who takes music? Investigating access to high school music as a function of social and school factors. Dissertation Abstracts International, 52(10), 3554A.

40.

Wolfe

E. E.

(1969). Relationships between selected factors and participation and non-participation in instrumental music in the Cincinnati Public Schools. Dissertation Abstracts International, 30, 2565A–2566A.

41.

Zervoudakes

Tanur

J. M.

(1994). Gender and musical instruments: Winds of change? Journal of Research in Music Education, 42, 58–67. doi:10.2307/3345338

42.

Zwick

Grief Green

(2007). New perspectives on the correlation of SAT scores, high school grades, and socioeconomic factors. Journal of Educational Measurement, 44, 23–45. doi:10.1111/j.1745-3984.2007.00025.x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.38 MB

0.00 MB