The Georgia Department of Education administers the annual Georgia Student Health Survey, which includes both questions about school climate and student health behaviors. Elementary students in grades 3-5 take a 15 question survey, and middle and high school students take a 91 question survey. Schools are required to have at least a 75% response rate.
The graph below shows how middle and high school students responded to the question, “I feel my school has high standards for achievement” during the 2018-2019 school year. Atlanta Classical scored the highest in the district on this question, followed by KIPP STRIVE. Mouse over or tap a result to see percentages for each response. 74% of students at Classical strongly agreed to the question and only 2% of students strongly disagreed.
Use the filters to view questions that interest you, change school levels, or view other metro Atlanta districts. This visual includes questions that focus on school climate, as opposed to student health behaviors. To view all results, see the state results page here.
The next graph shows elementary school responses to the question, “How often in the past couple months have kids picked on you by threatening you.” Survey results show that this was most common at Thomasville Elementary, where 11% of students selected “Every day” and 47% selected “Never”. Scroll down to see that this was least common at Springdale Park, where 1% selected “Every day” and 84% selected “Never”.
The next view shows the distribution of results for each question. The default view highlights results at KIPP Vision Primary. Students at this school report low levels of bullying (the questions starting with “How often”) and report high frequencies for positive questions like, “Good behavior is noticed at my school” and “Teachers treat me with respect.” Use the school filter on the right to change schools.
The secondary climate survey is 91 questions and anonymous. A potential concern with the data is that some students might not take the survey with validity, in which case the results are less useful. A way to investigate this concern is to correlate the climate results against an independent measure of school quality. If the measures are correlated, this is evidence that the climate survey is actually measuring a positive school climate.
As an external validity check on the climate survey results, we examined the relationship between climate survey results and the progress component of the CCRPI, which is based on student growth percentiles. Results of our analysis showed that most climate questions have a statistically significant relationship with CCRPI progress points when compared individually1. Many of the climate questions are correlated to other questions, so we also ran a combined model to see which questions are most strongly related to progress. When all questions were included in the same regression model, the questions we highlighted in the first two graphs above were the strongest predictors for progress: responding “strongly agree” to “I feel my school has high standards for achievement” at the secondary level, and responding, “never” to “How often in the past couple months have kids picked on you by threatening you” at the elementary level2.
The only question with an ambiguous relationship to CCRPI progress points at the elementary level when ran as an individual regression was, “I feel like I do well in school.” This is likely because the question is capturing two competing factors: whether the school has a supportive environment (which is good) and whether the school has low standards for academic achievement (which is bad). Hence the overall relationship with academic growth is ambiguous.
At the secondary level, some of the bullying questions do not have a statistically significant relationship to student progress when ran as individual regressions. Questions that measure similar issues but have a stronger relationship to progress are, “Students at my school fight a lot” and “I have been concerned about my physical safety at school.” These questions are likely better correlated with CCRPI progress because they describe more general events or concerns, instead of specific events that will be relatively rare.
To get email alerts for similar posts, please subscribe to the blog.
- Specifically, we ran a regression for each climate survey question, where 2019 CCRPI progress was the dependent variable, and also controlled for school district fixed effects and school demographics like percentage direct certification (A measure of school poverty levels). Control variables were selected based on whether they improved adjusted R-squared. In addition to APS results, we included Clayton, Cobb, Dekalb, Fulton, and Gwinnet for a larger sample size and increased statistical significance.
- To include all questions in the same model, we used a LASSO regression to address colinearity issues with similar climate questions. The two questions we list were the best predictors when using the full sample of schools and also when repeatedly running the model with a 75% random sample.