Suspensions & Expulsions at California Schools: How do they affect marginalized students?

DataRes at UCLA
10 min readMar 15, 2024

Kaera Mishenkov Mitchen (Project Lead), Sahil Nale, Srivarsha Rayasam, Junze He


Suspending K-12 students–is it a necessary punishment for bad behavior or is it a practice that perpetuates racial and class differences among youth? California lawmakers appear to concede that disciplinary action in public schools is an area in need of reform. Just last October, Gov. Gavin Newsom signed a bill into law enacting a ban on “willful defiance” (defined as “disrupting school activities or otherwise willfully defying the valid authority of school staff”) and truancy suspensions for middle and high schools. Proponents of this measure claim that suspensions and expulsions disproportionately harm minority students, who face more and lengthier consequences from disciplinary action compared to their peers. “Willful defiance” has been an especially controversial rule due to its vague definition — “disrupting school activities or otherwise willfully defying the valid authority of school staff.” This bill goes into effect in July, but these kinds of suspensions have continued to be implemented among most school districts in California — aside from several major school districts and elementary schools. We can identify features or trends about suspensions and expulsions across California through the lens of data to gain insight into this issue.

Over the course of several weeks, our DataBlog team analyzed public data provided by the California Department of Education to gain insight on educational equity as related to negative behavioral outcomes. Specifically, we assess whether differences in reports of expulsions and suspensions are associated with socioeconomic factors within school districts. We examine whether there are significant differences between districts and demographic reporting categories. Due to scope and time constraints, we focus on datasets from the 2022–2023 school year to give us the best representation of disciplinary stats as it pertains to the current educational climate.

Exploring Expulsion & Suspension Data:

The relative frequency graph above is the number of the total suspended students from different demographic designations grouped by each district in California. The density curve is not a bell shape, which means the number of the total suspended students from each race is not normally distributed and has a notable standard deviation. It shows that the number of total suspended students varies for different demographic reporting categories. This also implies that there are more negative outcomes in some categories than others.

By taking the number of suspensions/expulsions and dividing it by the student population of a district, we get a ratio to represent the rate of expulsions per district.

Looking at California as a whole, we find that suspension and expulsion rates do not appear to be proportional to the student population of a district. Expulsions are much rarer than suspensions and show minimal regional trends. Suspension rates vary across districts, but noticeably seem to be higher in more rural districts.

Upon examining our top 10 districts with the highest suspension/expulsion ratios, we confirm that there appears to be a significant concentration of disciplinary measures taken against students in rural districts; such as the inland empire, outer San Bernardino, and the far North of California.

Looking at the counts of total (all categories) and ‘defiance’ type suspensions by some of our key racial and socioeconomic categories, we see that defiance suspensions only make up a small amount of total suspensions. Note that an individual counted in the socioeconomically disadvantaged category can also be counted in the racial/gender demographic categories.

We examine the differences in reports between demographic categories in further detail by calculating the proportion of defiance suspensions represented by each of our key categories. In other words, we take the number of defiance suspensions for a category and divide it by our general total for defiance suspensions to obtain a proportion. Based on our graph, slightly over 80% of defiance suspensions involve students who are considered socioeconomically disadvantaged, which is a strikingly large proportion. According to the California Dept. of Education, students are reported as socioeconomically disadvantaged if they meet at least one of several criteria, including having parents who have not completed high school, being eligible for the Free or Reduced-Price Meal program, being considered homeless, etc. Students with these disadvantages may face even worse long term consequences in terms of economic mobility from these school disciplinary measures.

In regards to defiance suspensions for our key racial categories, the proportions appear to be close to the racial/ethnic distributions of California students as a whole, with some slight variation. Defiance suspensions are accounted for by 60% of hispanic/latino students and, according to our same CA Dept. of Education data source, 56.1% of the total student body of California public school students are hispanic/latino.

Relation to College Going Rate:

The relative frequency graph above shows the college going rate in 12 months (2023) after each student demographic had negative outcomes in 2022. The density curve of the graph appears to be a bell shape, which means the college going rate from different students with negative outcomes did not differentiate that much from each other nor have a large standard deviation. Even though some types of students were reported in 2022, they still had a nearly 25% chance of getting accepted by a college in 2023. Since the graph has a longer tail that is skewed to the left, it means some students had a lower chance of going to college based on their identity.

To further investigate the relationship between the number of suspensions for each category of reported student (2022), and their respective college going rate (2023) we grouped the suspension and college going rate dataset by districts and joined the two datasets together by the demographic reporting category. Then, we calculated the number of total suspended students and the average college going rate for each student type category (by race, gender, etc).

The students were categorized by gender, race, and other groups. The gender group generalizes students into 3 different types. The black square is an overall average college going rate across different students in all three groups, so we used the black square as a metric which helped us evaluate the college going rate from each type of student.

We could not find a direct correlation between total suspensions and demographic reporting categories. For example, Asian students only had 50 total suspensions but also displayed the highest college going rate in 2023 at approximately 34%. On the other side of the spectrum, non-binary students had nearly 0 suspensions and the lowest reported college going rate at almost 0%. It appears that race, gender, or other disadvantage/disability alone is more relevant to determining college going rate than the combination with the suspension count variable.

Non-binary students, migrants, and American Indians seemed to be outlier categories in our dataset. In gender, female students outperform male students in terms of college going rate and account for less suspensions. In the college going rate of race groups, Asian students had the highest college going rate, above the total average. Furthermore, many students in the other groups, such as students with disabilities, English learners and so on, were all below the overall average college going rate.

Is there a relationship between student populations with high expulsion rates and rates at which students attend college? Does poor student outcome relate to how many times students have been expelled from high school? Overall, there doesn’t seem to be any sort of relation between the number of students in a district who are enrolled in college and the numbers at which they’ve been expelled.

However, upon closer examination, the more expelled students in a district, the less students enrolled in California Community Colleges.

In a similar context, many students in populations with high expulsions have low UC school enrollment.

In another context, the number of students who go to college is not related to the total expulsion rate, as expulsion can still mean graduating from a different high school and enrolling in a CCC. This may indicate that although students are punished through expulsion, they may still be successful through enrolling in college because the nature of their defiance does not correspond with their emphasis on education.

Relation to Income:

The graph below shows us a visualization of mean income of different school districts by using zip code to find income. As expected, we can see that urban areas have larger mean incomes than our rural districts and we can also see certain zip codes are more affluent than others.

The graph below shows us a visualization of mean income of different school districts and college going rates. The mean income is depicted by the size of the dots and college going rates are depicted by color of them. We can see that higher income districts that have bigger markers also seem to be darker depicting a higher college going rate.

Larger cities have higher income and also appear to have higher college going rates. Once again, we see rural concentration in the lower end of income and college going rates. This suggests that in our earlier identification of rural school districts being hotspots for school suspensions, income may play a factor into why these communities see higher rates of disobedience and school discipline.

Through another chart shown below, we confirm there appears to be a negative between income and the number of student expulsions.

We can use statistics to assess the difference and its significance between the college going rates between high and low income school districts. We classified districts as such: school districts in zip codes with a mean income greater than the California average are considered high income, and those equal or less than the total average are considered low income. Because the dataset is large, we take multiple random samples and plot the distribution of these t-tests for our estimation.

We get the 95% confidence interval of -11.278635 and -5.1733 for the mean difference between low and high income student college going rates and since the interval is completely negative we can say that we are 95% sure that high income students have higher college going rates compared to low income students.

Final Thoughts:

Based on our findings, we see that disciplinary actions seem to affect socioeconomically disadvantaged students, including those in low income communities, at a high rate. We were unable to conclude whether certain racial groups are disproportionately affected by disciplinary actions. This does not necessarily mean that there is no relationship between race and suspensions, just that our data does not reflect these effects. Considering other factors that are not captured in this data, such as student grades, well-being, employment rates, among other factors that measure student outcome or “success/failure” in real life context may be a greater indicator of how suspensions affect students of different racial groups. Overall, our findings seem to be in favor of the “willful-defiance” suspension ban. Public schools, especially middle schools, are designed to mold teenagers into responsible adults and imbue them with skills necessary to graduate and enter the workforce (or college). Punishing students for misbehavior or disrespect of authority by removing them from the environment designed to teach them discipline seems counterproductive and detrimental in the long run, especially for socioeconomically disadvantaged students who may not have the financial freedom or outside support necessary to make up for gaps in their education.