Online learning for students in grades K-12 has been available for over a decade in some states, and the number of programs offered by school systems or virtual (online)1 charter schools is increasing rapidly. Connections Education has been operating Connections Academy online schools since the fall of 2002 and currently enrolls almost 50,000 students.2 Families seek out online schools for many reasons. Forty percent of the students who enrolled in Connections Academy schools for the 2012-2013 school year reported they needed more flexibility. Another 26 percent reported they did not get enough attention from their previous teacher. Fifteen percent reported they were not safe at school, and 12 percent indicated there was a health problem that limited a student’s ability to attend a traditional school. Approximately 33 percent indicated they have not been successful academically. One of the major criticisms of full-time online schools is that their state standardized test results are lower than students’ scores in traditional schools. In addition, while there are some full-time online programs that score well on some states’ accountability measures, most do not. We believe that this reflects a simplistic view of a complex issue. We believe that the value of these programs needs to be determined based on an analysis of the performance of comparable student populations in traditional schools and that the composition of the tested population also is important when comparing the performance of online schools to each other.
As a leader in the delivery of online education, Connections Education continually reviews internal data to improve student performance at the schools we serve. We have noted for some time that while many of the students in a Connections Academy school remain advanced or proficient on state standardized tests or improve their performance as compared to their prior results, others fail to improve or even show a decline in their performance. We needed a better understanding of the factors that contribute to these different outcomes in order to improve our performance and better understand the challenges presented by current measurement systems. Our analysis of standardized test scores for students enrolled in the schools we operate highlights several factors we believe may be adversely affecting the performance of online schools (as well as the performance of some traditional schools), which may need to be specifically addressed through different educational strategies and different accountability measures. For example:
- The timing of a student’s enrollment in a full-time online school impacts the student’s state test score performance, and a significant number of students enroll in online schools after the start of the school year.
- A statistically significant relationship exists between a student’s household income level and state test performance. Students who enroll after the start of the school year are more likely to be from lower income families.
- Family engagement with the decision to enroll in an online school has a positive influence on academic performance. Engagement during the enrollment process declines after the start of the school year.
Most states require online schools to accept students throughout the school year, unless they are subject to enrollment caps. There is increased demand for online schooling after the start of the school year as families look for solutions to problems that develop in their previous schools. About 30 percent of the new students we serve enroll after the start of the school year.3 Our data indicates that these students are less likely to score at the proficient level or above on their state’s standardized tests.4
Percentage of new students enrolled in a Connections Academy-operated school who score at the proficient level or above on their state’s standardized tests, administered during the spring of 2012, based on their time of enrollment (number of days after the scheduled start of their school)
Most states recognize that students who start after October 1 are less likely to do well on a state test given the limited time between their enrollment and the time of the test. So, they exclude students who enroll after October 1 from their accountability measures. But our data shows that a significant decline in student performance occurs well before an October 1 cut‐off date, most likely due to the conditions that caused the students to leave their previous school in the midst of the school year. With the exception of some traditional public schools that serve a highly transient student population, it would be unusual for a traditional school to enroll so many new students after the start of the school year. Therefore, online schools are disproportionately affected by the inclusion of the scores for students who enroll after the start of the school year but before a state’s cut‐off date.
It is well known that low-income students generally score lower on state standardized tests.5 Since schools that serve low-income students vary considerably in their access to highly qualified teachers, time devoted to instruction, curriculum materials, etc., it has been difficult to determine if household income produces students less likely to test well or if students from lower income households do not have access to the same educational resources as higher income students, or both. Full-time online school programs use the same teacher pool, curriculum and technology and set the same time requirements for instruction for all students, regardless of their income level. So comparing low-income students to higher income students within the same online school narrows the comparison to income, without the influence of different school experiences. As shown by reported test data from the South Carolina Connections Academy (SCCA) during 20126, there is a significant disparity in academic performance based on income level. 7
Percentage of all students enrolled in the South Carolina Connections Academy school who scored proficient or above on their state’s standardized tests, administered during the spring of 2012, based on their household income (as reported by the South Carolina Department of Education state web site)
Further, as shown below, the higher the percentage of low‐income students who are tested in the third grade, in all of the large full‐time virtual schools in Pennsylvania, the lower the percentage of their students who are proficient or above in math.
Percentage of all students in the largest “cyber” schools in Pennsylvania who scored proficient or above in third grade math on their state’s standardized tests, administered during the spring of 2012 (as reported by the Pennsylvania Department of Education state website)
The influence of income on the performance of virtual schools also is evident in Florida. The three full-time virtual schools that received that state’s highest rating of “A” for the 2012‐2013 school year had much smaller percentages of low income students than the program that received a “C”.8
Full-time public virtual schools almost always are required to accept all students, even though it is widely acknowledged that online learning does not work for everyone.9 In our effort to understand the factors that contribute to student success, we examined the level of engagement by a family with the school during the enrollment process.10 Our data shows that students who enroll later in the school year are less likely to attend an orientation event or have other interactions with the school (e.g. talking to a teacher or another parent) during the enrollment process. For example, 78 percent of families who start a Connections Academy school in August demonstrate engagement with the school during the enrollment process versus only 47 percent of those who enroll in December.
We are developing strategies to address all of these findings. Starting with the 2013-2014 school year — we have completely redesigned the “onboarding” process for students who start after the first day of school to provide them with more forms of support earlier, we are increasing our early warning systems to identify students at academic risk, and we will specifically include household income as an identified risk factor. We also are increasing enrollment engagement activities after the start of the school year and, to the extent permitted by school regulatory authorities, are requiring some level of engagement with the school by the parent or student prior to enrollment.
We believe that a better understanding of the factors that influence academic performance in online schools will help in the design of more effective evaluation systems for all schools. This is important because it is increasingly clear that many students want to learn online and many of them are doing so successfully. Our hope in sharing our data is that it will assist in this effort, particularly when determining which students and schools should be compared to each other and how to measure the impact of online learning as compared to traditional models for comparable populations.
1 “Online” and “virtual” are used interchangeably to refer to schools where the students and teachers are not in the same location and students use the Internet in order to access instruction and curriculum materials.
2 Connections Education serves students in full-time online programs that are operated under contracts with charter schools, school districts and departments of education in 22 states.
3 Full-time online schools follow a traditional school calendar because their students must take their state’s standardized tests and the tests are administered to all public school students at the same time. While there are discussions about decoupling testing and funding models from the traditional school year, this is not yet practical for schools that are measured under existing state accountability systems.
4 While states currently use different standardized tests, they all report the percentage of students who meet or exceed their designated proficiency threshold, so we have combined the data from multiple Connections Academy schools on that basis.
6 SCCA tests students at Grades 3-8 and students in their second year of high school. In South Carolina, Grades 3‐8 take the PASS and 2nd year high school students take the HSAP. Values in the table are based on results calculated across PASS and HSAP results for SCCA.
7 South Carolina is one of the few states that report the performance of students who are not from low-income households as well as those who are. So in most states, it is impossible to see the difference in performance on state tests solely based on income. Yet all states report detailed comparisons of test score performance based on ethnicity.
8 Broward Virtual, school grade of A with 20% low-income students reported; Palm Beach Virtual, school grade of A with 18% low-income; Lee County Virtual, school grade of A with 33% low-‐income (but requires low scoring students on the FCAT to attend in-person remediation sessions, which potentially disqualifies it for analysis as a completely virtual school program); versus Florida Virtual School Full-Time with a school grade of C and 48% low-income reported. http://schoolgrades.fldoe.org/
9 “Elearning works, Exactly how well depends on its unique features and barriers”, January 2013, Cornell University. While this report covers higher education, the issues are the same in online learning in K-12.