the individual level. Level 2 investigated how student behaviors in 2010 can be predicted by school size, the percentage of minority students, the total number of school policing, and the percentage of economically disadvantaged student. The equation format at level two for student behaviors involving insubordination and violence is as follows:
๐'# = ๐ฝ''+ ๐ฝ'1 ๐๐๐ง๐# + ๐ฝ'5(๐๐๐๐๐๐๐ก๐ฆ") + ๐ฝ'= ๐๐๐๐ข๐๐๐ก๐ฆ" + ๐ฝ'@ ๐๐ธ๐" + + ๐'" Where ๐ฝ'' is the expected average number of student behaviors for participating schools at the initial status. The regression coefficients, ๐ฝ'1, ๐ฝ'5, ๐ฝ'=, and ๐ฝ'@ , represent the strength of the relationship between student behaviors and predictor variables including
school size, percentage minority, the total number of school security, and the percentage of economically disadvantaged students, respectively.
Likewise, individual change modeling at Level 2 was also used to determine the rate of change in the average number of student behaviors over a four-year time period as related to school size, the percentage of minority students, the total number of school policing, and the percentage of economically disadvantaged students. This equation is represented below.
๐'# = ๐ฝ1'+ ๐ฝ11 ๐๐๐ง๐# + ๐ฝ15(๐๐๐๐๐๐๐ก๐ฆ") + ๐ฝ1= ๐๐๐๐ข๐๐๐ก๐ฆ" + ๐ฝ1@ ๐๐ธ๐" + + ๐1" Where, ๐ฝ1' is the estimated intercept of the expected rate of the change in the average number of student behaviors for a typical school. The coefficients, ๐ฝ11 , ๐ฝ15 , ๐ฝ1=, and ๐ฝ1@, are the contribution of school size, the percentage of minority students, the total number of school policing, and the percentage of economically disadvantaged students to the rate of change in student behaviors by school building. ๐1" represents the random error associate with the rate of change for school ๐. Assumptions associated with this model include independent and normal distribution with a mean of 0.
Summary
Chapter III described the research methodology, including the use of individual change modeling to determine the impact of school security measures on student behaviors involving insubordination and violence over time. The chapter also detailed a pilot study that was used to inform this dissertation.
CHAPTER IV FINDINGS Introduction
Chapters I through III introduced the study and the theoretical framework, a review of the literature, and the research methodology. Chapter IV will provide a
descriptive analysis of the characteristics of the participating schools and the results from an individual change model. This analyses focused on the extent to which the presence of school security measures involving SROs and police officers impact student behaviors that include: insubordination and violence, at the initial status (2010) and over time (2010-2014). Other variables of interest included the percentage of students identified as economically disadvantage, the percentage minority, and school size.
Data collection for the current study included a survey that was disseminated to principals in Ohio public schools in suburban, rural, and urban locales as well as towns via SurveyMonkey. Specifically, principals were asked the following questions: 1. the total number of SROs present in their schools and 2. the total number of police officers present in their schools. Incidents of student behaviors involving insubordination and violence were obtained from the ODE (2014) website. This study had a low response rate of 6.5% as 2,583 principals in the state of Ohio were emailed the survey and, of
these, 167 principals responded. As mentioned, school principals were asked to indicate the total number of police officers and security officers that were on duty in their schools while the student behavior data was obtained from the ODE (2014) website. Schools that did not appear in the ODE dataset were excluded from the study. Also, a number of principals did not provide their school and/or their schoolsโ district, making it impossible to determine the number of student behaviors for these schools. In the end, of the 167 principals that participated in the study, data from 148 of these schools were able to be examined in the study.
As previously discussed, this dissertation was informed by a pilot study, which was analyzed using an individual change model. This was an appropriate model to analyze these data as individual change modeling can be used in longitudinal studies to determine the impact of a given predictor on an outcome variable of interest over time. In the present study, an individual change model was used to make repeated observations of student behaviors involving insubordination and violence at the individual school level with respect to the presence of school security measures. Additional variables of interest were also analyzed over the provided four-year time span using this model and included school poverty, school size, and percent minority for each observed school. The total number of school security guards were controlled for every 1,000 students. Although this study was informed by the aforementioned pilot study and many of the procedures
remained the same in this dissertation, some changes were made to the proposed methods in the present study due to the number of participants and the characteristics of the
In chapter III, the proposed method for disaggregating the data involved using the typology categories; however, due to the low response rate, it was not possible to analyze the data in this manner. Instead, all school locales were aggregated into one outcome variable. It was also proposed that consideration to school level (e.g. elementary) should be given; however, since the sample size was low, it was also not possible to disaggregate the data in this manner. Therefore, elementary, middle, and high schools were all
examined together with respect to the total number of school security measures. Another notable difference between the pilot study and this dissertation was the fact that the pilot study asked principals to indicate the total number of security guards present in their schools, while the current study asked principals to distinguish between the total number of SROs and police officers in their schools. This distinction was made due to the literature presented by Nolan (2011) that discusses these job titles separately. However, upon examination, these variables were highly correlated and were merged together to form one predictor variable, denoted as school security measures in the
present study. In addition, rather than adhering to the same process used in the pilot study involving disaggregating data by high and low poverty, this study examined the
percentage of economically disadvantaged students in order to garner a more accurate data picture regarding the impact of poverty of studentsโ behavioral outcomes. Schools that are identified as high or low poverty may differ in the percentage of students from economically disadvantage households within these distinct categories.
The next section will provide a detailed description of the characteristics of the data, followed by the findings of an individual change model used to determine the extent to which student behaviors involving insubordination and violence are related to school
security measures per every 1,000 students at the initial status (2010) and over time (2010-2014). Findings also consider how these behaviors relate to school size, percentage minority, and percentage of schools that are economically disadvantaged. Conclusions regarding these findings and a summary of the chapter are also included.
Descriptive Characteristics
The first step in analyzing the data was to determine the school characteristics of the sample. The mean school size was 538 students with a range between 108 and 2331. The mean for the percentage of minority populations in the sample was 14.98. This number is representative of minority groups in Ohio as only 12.6 percent of Ohioโs population is comprised of Black Americans (U.S. Census Bureau, 2014). The
breakdown by Typology included schools in rural, towns, suburban, and urban settings. Twelve percent of the participating schools were identified as being in high poverty and 17 percent were considered low poverty rural areas. Fourteen percent were located in high poverty and 12 percent were located in low poverty towns. Eleven percent of participating schools were located in high poverty and 21 percent low student poverty suburban settings. Nine percent were considered urban schools with high and 5 percent very high student poverty rates. As mentioned, due to the low sample size, school typology was only used to understand the descriptive characteristics of the sample. The average percentage of students identified as being economically disadvantaged in the sample was 42.08.
Findings of the Study
The total number of student behaviors for each participating school for the four years of interest, between 2010-2014, with respect to the school size, percentage
minority, percentage economically disadvantaged, and the presence of school security measures for the initial status and the rate of change were of interest in the present study. An individual change model (Williams & Raudenbush, 1989) was used to determine the relationship between student behaviors involving insubordination and violence and the presence of school security measures, percentage minority, percentage economically disadvantaged, and school size.
To answer questions involving the impact of these predictor variables on student behaviors at the initial status and over time, an individual change model was used. The variables used in this analysis are the total number of school security guards per every 1,000 students and the total number of student behaviors involving insubordination and violence, all of which are continuous. Other variables of interest included 1. size, 2. percentage minority, and 3. percent economically disadvantage. Findings with respect to student behaviors involving insubordination and violence appear below.
Research questions #1.
To what extent do studentsโ incidents of insubordination in 2010 (initial status) relate to the presence of school security measures in Ohioโs public schools?
Table 1 presents findings of an individual change model at the initial status (2010) for insubordination.
Table 1.
Summary of an Individual Change Model regarding the extent to which studentsโ incidents of insubordination in 2010 (Level 1).
B ๐. ๐ธ. ๐ Level 1 Minority -.24 .44 .60 Size .20 .04 .00 Security 61.36 12.68 .00 SES 1.36 .44 .01
For the dependent variable, insubordination, the individual change model revealed statistically significant findings for school size, school security, and the percentage of economically disadvantaged students at the initial status (๐ = .20, ๐ = .001; ๐ = 61.36, ๐ = .001; and ๐ = 1.36, ๐ = 0.01 respectively). As school size, the total number of school security measures, and the percentage of economically disadvantaged students increased, so did incidents of insubordination in 2010. The percentage of minority students was not found to predict insubordination at the initial status.
Research questions #2.
To what extent does the rate of change in studentsโ incidents of insubordination between the 2010-2011 and 2013-2014 school years relate to the presence of school security measures in Ohioโs public schools?
Table 2 presents results from an individual change model for the rate of change in studentsโ incidents of insubordination between 2010 and 2014 relates to the presence of school security guards in Ohioโs public schools.
Table 2.
Summary of Individual Change Modeling involving how the rate of change in studentsโ incidents of insubordination between 2010 and 2014 (Level 2) relates to the presence of school security guards in Ohioโs public schools.
Level 2 B ๐. ๐ธ. ๐
Minority .00 .11 .75
Size .00 .01 .99
Security -.08 3.31 .59
SES 1.80 .12 .46
The predictor variables were not statistically significant for the rate of change in student behaviors involving insubordination during the examined four-year period (2010- 2014).
Research question #3.
To what extent do studentsโ incidents of violence in 2010 (initial status) relate to the presence of school security guards in Ohioโs public schools?
Table 3 presents findings of an individual change model at the initial status (2010) for violence.
Table 3.
Summary of Individual Change Model regarding the extent to which studentsโ incidents of violence in 2010 (Level 1) relate to the total number of school security measures.
B ๐. ๐ธ. ๐ Level 1 Minority -.02 .05 .72 Size .04 .00 .00 Security 4.73 1.60 .04 SES .13 .06 .11
For the dependent variable, violence, the individual change model revealed no statistically significant findings for the percentage minority or the percentage of economic disadvantage. Statistically significant findings were revealed for school size (๐ = .04, ๐ = .001) and the presence of school security (๐ = 4.73 ๐ =
.04, respectively). As school size and the total number of school security measures increased, so did violence in 2010.
Research question #4.
To what extent does the rate of change in studentsโ incidents of violence between the 2010-2011 and 2013-2014 relate to the presence of school security guards in Ohioโs public schools?
Table 4 presents results from an individual change model for the rate of change in studentsโ incidents of violence between 2010 and 2014 relates to the presence of school security guards in Ohioโs public schools.
Table 4.
Summary of Individual Change Modeling regarding how the rate of change in studentsโ incidents of violence between 2010 and 2014 (Level 2) relate to the presence of school security guards in Ohioโs public schools.
Level 2 B ๐. ๐ธ. ๐
Minority .00 .03 .90
Size -.00 .00 .41
Security -.45 .87 .61
SES .01 .03 .60
The predictor variables were not statistically significant for the rate of change in student behaviors involving violence during the four-year period (2010-2014).
Overview
There were four research questions that guided this study including: 1. To what extent do studentsโ incidents of violence in 2010 (initial status) relate to the presence of school security guards in Ohioโs public schools? 2. To what extent does the rate of change in studentsโ incidents of violence between the 2010-2011 and 2013-2014 relate to the presence of school security guards in Ohioโs public schools? 3. To what extent do studentsโ incidents of violence in 2010 (initial status) relate to the presence of school security guards in Ohioโs public schools? 4. To what extent does the rate of change in studentsโ incidents of violence between the 2010-2011 and 2013-2014 relate to the presence of school security guards in Ohioโs public schools?
The data for the present study was examined using an individual change model. The presence of school security measures increased problematic student behaviors involving insubordination at the initial status (2010), but not over time (2010-2014). The presence of school security measures also increased student incidents of violence at the initial status but was not statistically significantly related to these behaviors over time. This finding indicates that school security measures did not effectively reduce violence within the examined schools.
The percentage of students who are economically depressed was statistically significantly related to the increase of student behaviors involving insubordination but was not found to be statistically significant for violence or over time for the examined behaviors. School size was statistically significantly related to problematic student behaviors involving insubordination and violence at the initial status only. Findings also revealed that the percentage of minorities present in Ohioโs public schools was not linked to high rates of problematic student behaviors at the initial status or the four-year period (2010- 2014). The following will focus on conclusions regarding these findings
Summary
Chapter IV revealed the findings of the study as well as the data analysis. This chapter included an analyses of the extent to which student behaviors involving insubordination and violence are related to the presence of school security, percentage poverty, percentage economically disadvantaged, and the percentage minority at the initial status (2010) and over time (2010-2014). Findings showed that school size, socioeconomic status, and school security were statistically significant predictors for both insubordination and violence at the initial status. There were no statistically
variables. The next chapter will focus on the conclusions of the study and includes a discussion and recommendations for future research regarding school discipline in American public schools.
CHAPTER V
DISCUSSIONS, CONCLUSIONS, AND RECOMMENDATIONS