**8. Parents' perceptions of school inclusiveness**

A total of 619 parents (51.37% in the treatment group and 48.63% in the control group) provided valid responses to the parent SIS administered in May 2020. Controlling for grade levels, no statistically significant differences were found between the two groups of parents in any of the three aspects (i.e., inclusiveness, positive relationship, and fairness). However, qualitative analysis of parents' empathy mapping showed that parents felt their voices diminished calling concern to safety, security, and sense of care and belonging for students and families of diverse backgrounds. The results of the analysis also indicated poor accessibility to information and material resources for parents of limited and non-English-speaking communities. While we were unable to analyze parent surveys, we also conducted qualitative interviews to gather insight on how if they felt welcomed when they entered their schools.

## **9. Teacher outcomes**

Teachers' chronic absenteeism is defined as missing more than 10 days (5.5%) in a typical 180-day school year. At baseline, 22.90% of teachers in the control group and 25.48% in the intervention group were chronically absent. In 2019, 20.84% of teachers

<sup>1</sup> Due to the smaller sample size of leaders, we used 0.1 as the significance level in the hypothesis testing.


#### **Table 5.**

*Impact estimates on student outcomes for elementary and middle schools.*


#### **Table 6.**

*Impact estimates on student outcomes for high schools*

in the control group and 21.29% of teachers in the intervention group were chronically absent. There was a 2.06% drop in the rate of chronic absenteeism in the control condition and a 4.19% drop in the treatment condition. However, the two were not statistically significantly different as indicated by an HLM analysis.

*School Improvement Inclusion Model for Schools with Changing Demographics: The Impact... DOI: http://dx.doi.org/10.5772/intechopen.113114*

## **10. Student outcomes**

To examine the impact of the school improvement model on student outcomes, we examined whether there are differences between the intervention and the control conditions. These included student chronic absence, suspension, math achievement, reading achievement, and college and career readiness. Because the study uses a quasi-experimental design, there were systematic biases between the students in the intervention condition and those in the control condition. Intervention students were more disadvantaged than the control group. Thus, students in the intervention group dealt with more issues of poverty, language barriers, and parent engagement. Hence, students in the intervention group were more likely to be English language learners (ELLs), less likely to be economically advantaged, and had lower baseline math and reading test scores. To reduce the effect of these biases on the impact estimates, we used propensity score analysis (PSA), in which students were matched on baseline measures of achievement as well as grade levels and demographic characteristics including gender, race, ELL status, special education status, free or reduced lunch eligibility, and gifted program status.

In the last year of the grant, the COVID-19 really influenced the work with students in the treatment group. It is important to note that when we approached the district about participating in the project, the district assigned us the lower performing schools that had a similar feeder pattern. We believe if COVID-19 had not occurred, student outcomes may have improved.

Two sets of PSAs were conducted, one for elementary and middle schools (ESMS) and the other for high schools (HS). In both analyses, the matching successfully reduced biases in covariates to less than 4%. **Table 5** shows the results of the ESMS analysis, including sample size, grade levels, average treatment effect (ATE), statistical test results, and robustness. There was one statistically significant negative effect (i.e., chronic absence), one statistically significant positive effect (i.e., Algebra), and three nonsignificant effects (i.e., suspension, math STAAR test scores, and reading STAAR test scores). Specifically, treatment group students were more likely to be chronically absent by 2%. In addition, treatment group students had higher Algebra end-of-course (EOC) scores by 0.11 standard deviations. The robustness parameter measures how robust an estimated impact is against unmeasured confounders. Values closer to 1 indicate less robust effects. The estimated robustness parameters of the two statistically significant effects indicate that both effects were very sensitive to potential confounders.

The PSA for high schools only matched students on demographic characteristics because one high school in the treatment group did not enroll any students in the first year, thus did not have baseline measures of student achievement, attendance, and suspension. The results (see **Table 6**) showed one statistically significant positive effect, six statistically significant negative effects, and two nonsignificant effects. The high school results should be interpreted with caution. This was because there was only one high school in the treatment group and only one high school in the control condition. Thus, it was a challenge in separating school effects from the treatment effects. In addition, the lack of baseline measures also made the results less robust.

## **11. Results and application for schools**

Most reform efforts in addressing changing demographics are based on awareness training, equity audits, and culturally competent training for school leaders. While each method has noted some success, this inclusion model used multiple strategies.

Thus, this research explored the potential of developing a more holistic model to create inclusive schools for all demographic groups. An important element for implementing this model was the development of a survey to measure inclusion. Survey constructs measured the degree of responsiveness to students and parents. Thus, this instrument assessed the principals' cultural competence, culturally relevant strategies, and fair and equitable treatment for all students. Use of the survey allowed the researchers to measure if this school improvement model was effective in creating inclusive schools. The model was based on a wide array of strategies which focused on engaging demographically diverse parents, culturally competent training for school leaders, analyzing school equity data, implemented postsecondary strategies at the elementary and middle schools.

Interventions were developed to address the model's three dimensions of outcome awareness, organizational justice and fairness, and leadership capacity. Over a 3-year period, interventions were implemented to address inequities and engage parents. Theoretically, this study provided strategies to use with schools to improve student outcomes. It also included a survey based on inclusion constructs to assess if the schools were truly measuring and addressing inequities in school outcomes. Findings were positive which indicated this model has promise.

Findings conclude that the intervention had a statistically significant positive impact on teachers' and leaders' perceptions about creating inclusive schools. The qualitative data corroborated the statistical findings. However, the impact on student outcomes is indeterminate due to mixed effects and confounding factors. Based on the findings from this study, this improvement model has potential to assist schools in addressing a change in their student demographics.
