Title: Office referrals and suspension: Disciplinary intervention in middle schools
Authors: Skiba, Russell J. and Peterson, Reece L.
Source: Education & Treatment of Children; Aug 97, Vol. 20 Issue 3, p295, 21p, 5 charts, 2 graphs
Doc. Type: Article
Persistent link: http://search.epnet.com/direct.asp?an=195&db=f5h&tg=AN

This article summarizes a study that analyses the types of office referrals given in middle school and whether students of specific categories are disproportionately referred within school communities. The authors chose a large, urban Midwestern public school district, which is among the 15 largest cities in the United States and a medium sized Midwestern school system in a city population of approximately 200,000. The authors completed two studies within the subject; one more general study which focuses on the types of referrals made and the punishments incurred, and a more specific study of one school to analyze the referral/discipline system.

In the first study the authors completed a cross-sectional, deductive research study by collecting quantitative data from a school district, which serves over 50,000 students. They completed a stratified sample, evenly divided between sixth, seventh and eighth grade students, and four students who were listed as ninth graders. Of these students, 51.8% were male and 48.2% were female. These students were generally categorized as Caucasian (42%), African American (56%), Latino (1.2%), Asian American (0.7%), and Native American (0.1%). Within this sample, 83% of the students were in general education (16.8% special education) and 65.3% qualified for free or reduced lunch as an indicator of SES. The disciplinary records were received from the school system computer that tabulates the student’s information, triggering event and the resulting punitive action. If a student transferred out of district, was expelled or dropped out, they were removed from the study.

The authors compiled the data for the four independent variables within the student’s demographic data: gender, ethnic status, disability label and SES as established by free or reduced lunch eligibility. A one-way analysis of variance tested the dependent variables of either the number of referrals of number of suspensions. The analysis of the data, in regards to the number of referrals, found that there were significant racial differences between students referred to the office, with African American students receiving the most referrals, except Native American. It was also determined that students who receive free or reduced lunch, were male, or were labeled with a disability were also more likely to receive office referrals. When the data was evaluated in terms of the number of suspensions, it derived the mean number of suspensions per student to be low (0.5), with students receiving referrals averaging 1.25 suspensions per year. In regards to the independent variables, it found similar results to the other dependent variables mentioned above. African American students received the most suspensions, except for Native American students, and students who received free or reduced lunches and were labeled as disabled also received more suspensions. Boys also received more suspensions again. It was also found that when the data was analyzed in regards to the type of referrals, disruptive and misconduct behaviors were the most frequent types of offenses while more serious offenses such as weapons, fighting and sexual activity were less frequent. Conversely suspensions were the most frequently used method of punishment with alternative solutions less frequently used to correct the behavior.

In the second study, the authors attempt to itemize more detailed information about the specific types of offenses referred to the office and the types of punitive actions taken by the administrator. They chose a school 600 miles from the first study in a city with a population of approximately 200,000, and served about 6,770 students in the nine middle schools. The school’s population served 610 students in grade 7 – 9 and the minority population consisted of 9.2% African American, 1.5% Asian American, 0.8% Hispanic American and 0.5% Native American. 51.1% of the students were male (48.9% female) and approximately 29.5% of the population qualified for free or reduced lunches. The data was retrieved in the same manner as in the first study.

The authors found that the most common reason for office referrals were for disobedience and disrespect with more serious behaviors such as fighting, possessing weapons and sexual harassment/assault as more infrequent of office referrals. When the data was used to assess the characteristics of the students who received the referrals, 75.4% were male. Of the 38.2% of the school population who received office referrals, 47.1% were students eligible for free or reduced lunches and 38.6% were special education students. There was no mention of ethnic or racial differences. When the authors examined the actions taken by the administrator, sending a report home, other actions, and contacting the parent were the most frequently used course of actions, while suspensions remained relatively low on the list.

In many ways these studies used good research practices. In the first study they used a stratified random sampling strategy to correct for grade differences between six, seven and eighth graders. Anyone who has taught middle school knows there is a world of difference between each grade. If the sample were not stratified, your results may be skewed one way or another, depending on the weight of the grade level. I also found it interesting that this study repeated what other studies have shown, that certain student characteristics are more frequently referred to the office, specifically African American, low SES and students with disabilities, showing strength of association between the variables and good external validity measures for their research. The first study drew a sample from a large population and was able to use statistics that were already available within a database. This allows this study to be generalized to a larger population, nomothetic. The quantitative analysis of the data measured the variance of the dependent variables with the independent variable with much supported conclusions. This method limits the causal direction of the variables and supports a stronger relationship between the variables, or correlation. I have read other studies where many other variables are assessed through surveys of family life and social life.

In the second study, they also used good research practices, but they were not as good as the first study. They completed a deductive study using cross-sectional, quantitative data but are idiographic since they chose a single school. This data cannot as easily be generalized as the first study. They opted to study all the data for all the students in the school, 610 students in seventh, eighth and ninth grade. This is a non-probability sample, specifically purposive in which they are looking for an illustrative case. There are some difficulty in heterogeneity in that the school in study two did not contain a population that was as representative of minorities as the first study. This information could not be significantly compared to the first study, even though they did make these conclusions at times. Because of this population choice, there is not as strong a relationship between many of the dependent variable, specifically minority categories and SES. There is still a lot of strength in the casual direction of influence, but there is some more possibility of spurious variables due to the population chosen and the sampling strategy, but still is not severe since the first study supported many of the statistical conclusions.

Statistically, both studies used ordinal independent variables (number of referrals and number of types of consequences) with nominal dependent variables (gender, race, SES and disability label). In study one, they used a “One-way analysis of variance …to test for differences in the dependent variables of either number of referrals or number of suspensions by four independent variables drawn from the demographic data: gender, ethnic status, disability label, and eligibility for free or reduced cost lunch.” Although this was not covered in the statistics discussion, I continued my research on the Internet and I believe that this process is an elementary statistical process and may be a one-way ANOVA with categorical predictor variables. This way the authors looked at one of the variables as examined under several categories. They also showed the relationship to be statistically significant using a p-level, all at p<.001. To determine a statistical difference between the different groups within the variables, they completed a Newman-Keuls test, which is a test completed after the statistics are assessed to show that the difference is not just due to chance. In study two the authors simply categorized the type of each referral: reason for office referral, actions taken prior to the current offense, and action taken by the administrator. They then itemized the data within each category. There were no statistical procedures to compare the groups in this section of the study. They did evaluate the student data, as in study one but did not mention statistical processes used, but evaluated each group as a percent within the dependent variables.

Some statistical conclusions were not appropriately made. The authors compared the statistics from the first study and the second study. The first study took place in a very urban school system with a large minority population (58%) and low SES population (65.3%). This same population was not used in the second study of a single school where there was a small minority population (12%) and low SES population (29.5%). The statistics from both studies were compared in several conclusion statements, but because of their population differences, this comparison should not have been made and could lead to misconceptions about the data. This difference in the minority population may be a spurious variable between the two study outcomes, especially when in the second study race was not mentioned as a significant difference in referral rates. The increased SES and reduced minority numbers may have affected this data. The authors also compared different grade levels in the two studies. In the first study, they examined a middle school that contained sixth, seventh and eighth graders. In study two they examined a middle school that contained seventh, eighth and ninth graders. These students are at different developmental stages and may have different resulting behaviors.

The author’s ethical practices were generally good. They protected the students from harm because there was no involvement directly or indirectly with the students. Volunteering also was not an issue. They used data that was readily available within the school system computer. They received consent from the school system and through that consent any notifications would have been made to the public or administrators. The authors also protected the anonymity of the students and the school system by only generally referring to the cities chosen by statistical category and general geographic location. Since the article was published in a research journal, the findings were reviewed by others and were judged as accurate. This information was beneficial for the general public because it sheds light on the realities of office referrals and there was minimal or no risk to the public in regards to this study.

These studies identify some of the factors that middle school staffs face in the educational process. Middle school is a very difficult time for many students and an increase in office referrals often results in their struggle to come to terms with many physical, intellectual, and emotional factors in their lives. This often shows in a student’s inappropriate behaviors. Many of them work their way out of this but some never seem to come to terms with their conflicts. Today, in light of many current events, school discipline is a very hot topic. Studies have shown that behaviors have changed little over history and that the discipline process often deals with punishment over solution in many cases. This study is one of the first, in what should be many to follow, to help us understand what works and what does not.