CBM at Home - Shiny Contest Submission

CBM at Home

Authors: Yaacov Petscher , A.J. Torgesen, Marissa Suhr, Lillian Durán, Norma Medina,

Abstract: During the Covid-19 pandemic many children were not able to attend school in person. In person testing of early childhood reading levels was also limited. We developed CBM at Home as a way for parents to test their children's reading level and interpret the scores. We have also developed a fully Spanish version of the app, MBC en el Hogar.

Full Description: CBM at Home provides a way for parents to administer Curriculum Based Measures (CBM's) on their children from their homes and interpret the scores based on cut points derived from previous data.

When first navigating to the app the user is given the option to choose between the English or the Spanish version of the app. This will direct the user to one of two different shiny apps.

  1. After navigating to the language of your choice you will see the first screen of the app plays a video explaining what a CBM is and how to administer it to your child.

  2. The next page allows the parent or guardian to select the grade level of child being tested.

  3. The third page of the app will provide informative videos showing how to administer the CBM in the correct way. It will also explain how to input your child's score.

  4. The fourth page provides the CBM as well as the instructions for how to administer it. It also provides a printable pdf with the correct passage or letters for the child's use. This page uses javascript code to allow the parent to highlight the text that the child read and generates the word count for them, eliminating user error in counting of words. The parent then inputs the number of errors and the final score is calculated.

  5. The fifth and final page of the app shows the child's score in red or green indicating if the child is at risk for reading difficulties or not and a video to interpret that score. The page also provides additional resources for the parent to pursue should they wish.

Data were drawn from a large archival database of reading scores maintained at Florida State University. The full archival database included over 1.5 million records of students across grades K–10 in any given year from 2003 through 2014. Because the calibration process was reliant on a longitudinal participant pool, a sample of 60,403 students were identified across three longitudinal cohorts of students who maintained at least one record of data from K-3. Students were distributed across 1,077 schools, and 63 districts and were administered curriculum-based measures (CBMs) in K-3 and standardized end-of-year reading comprehension measures in grades 1-3.

Selected cut-points for each grade-level CBM that maximized the >= .80 sensitivity and >= .85 negative predictive power criteria was as as follows: Kindergarten (<26 letters correct per minute is at-risk), grade 1 (<12 words correct per minute is at-risk), grade 2 (<62 words correct per minute is at-risk), and grade 3 (<80 words correct per minute is at-risk). Lower specificity values were observed in each of the grades implying a higher false positive rate (i.e., Cell B in Table 1) that ranged from 30% in grade 1 to 60% in kindergarten. The high false positive rate in kindergarten was expected as only a single measure of child reading ability is used (i.e., Fall LNF) and the outcome is administered approximately 1.5 years later. Conversely, the negative predictive power and sensitivity values were all strong such that if a child is identified as not at-risk in CBM at Home there is a strong probability that the child will be successful on the measured outcome. It is this characteristic of the screener that we opted to maximize such that if a child is identified as not at-risk, there is a lower probability of missing out on additional reading support.

For a full breakdown of the technical specifications visit this link

We would like to thank all that helped work on this project including: Lakeisha Johnson, Sarah G. Sayko, Cy Stanley, Chris Stanley, Declan Stanley, and Lauren Stanley.

CBM at Home was supported by funding from the Chan Zuckerberg Initiative and the Office of Elementary and Secondary Education, in partnership with the Office of Special Education Programs (Award #: S283D160003). The opinions or policies expressed are those of the authors and do not represent views of CZI, OESE, OSEP, or the U.S. Department of Education.

Keywords: K-12, Education, Reading, Testing, Spanish, Bilingual
Shiny app: https://qmi-fcrr.shinyapps.io/CBMHome/
Repo: GitHub - AJTorgesen/CBMatHome: Bilingual App to test K-3 reading ability from home
RStudio Cloud: RStudio Cloud


Full image: