|
Field
Last Published
|
Before
July 15, 2024 04:17 PM
|
After
July 15, 2024 04:45 PM
|
|
Field
Experimental Design (Public)
|
Before
To assign parents to one of the texting groups, we will employ a multi-site person-level randomized controlled trial design, blocking on, for example, middle school site, grade level, and texting language, English or Spanish (Spybrook, Bloom, Congdon, Hill, Martinez, & Raudenbush, 2011). Specifically, within the districts’ middle schools, grades, and texting languages, we will randomly select half of participating parents to join the control texting group and half to join the INSPIRE-EXCEL-BOND texting group.
District 1 parents who participated in the pilot year program and who re-enrolled in 2019-2020 were assigned to their earlier treatment condition. This design allows us to examine whether multiple years of treatment has effects on student outcomes, relative to zero years of treatment.
|
After
To assign parents to one of the texting groups, we will employ a multi-site person-level randomized controlled trial design, blocking on, for example, middle school site, grade level, and texting language, English or Spanish (Spybrook et al., 2011). Specifically, within the districts’ middle schools, grades, and texting languages, we will randomly select half of participating parents to join the control texting group and half to join the INSPIRE-EXCEL-BOND texting group.
District 1 parents who participated in the pilot year program and who re-enrolled in 2019-2020 were reassigned to their initial treatment condition. This design allows us to examine whether multiple years of treatment has effects on student outcomes, relative to zero years of treatment.
|
|
Field
Planned Number of Clusters
|
Before
501 students in 2018-2019
1,156 students in 2019-2010
48 blocks in 2018-2019 (school-by-grade-by texting language fixed-effects)
42 blocks in 2019-2020 (school-by-grade-by texting language fixed-effects)
|
After
2018-2019 (same-year effects sample sizes)
District 1: 501 students, 48 blocks in 2018-2019 (school-by-grade-by texting language fixed-effects)
2019-2020 (same-year effects sample sizes)
District 1: 1,156 students, 42 blocks in 2019-2020 (school-by-grade-by texting language fixed-effects)
|
|
Field
Planned Number of Observations
|
Before
501 students in 2018-2019
1,156 students in 2019-2010
48 blocks in 2018-2019 (school-by-grade-by texting language fixed-effects)
42 blocks in 2019-2020 (school-by-grade-by texting language fixed-effects)
|
After
2018-2019 (same-year effects sample sizes)
District 1: 501 students, 48 blocks in 2018-2019 (school-by-grade-by texting language fixed-effects)
2019-2020 (same-year effects sample sizes)
District 1: 1,156 students, 42 blocks in 2019-2020 (school-by-grade-by texting language fixed-effects)
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
2018-2019
- Treatment: 253
- Control: 248
- Total: 501
- Blocks: 48
2019-2020
- Treatment: 581
- Control: 575
- Total: 1,156
- Blocks: 42
|
After
2018-2019 (same-year effects sample sizes)
District 1
- Treatment: 253
- Control: 248
- Total: 501
- Blocks: 48
2019-2020 (same-year effects sample sizes)
District 1
- Treatment: 581
- Control: 575
- Total: 1,156
- Blocks: 42
|
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
We used the PowerUp! tool for power analyses (Shiny App here: https://powerupr.shinyapps.io/index/), allocating approximately 50% of students to the treatment and 50% to the control. We assumed that our model would include 20 covariates and would explain about 33.33% of the variance. Below are the minimum detectable effect sizes (SD units).
2018-2019
PSJA: 48 blocks, harmonic mean across blocks = 5.937367 --> MDES = 0.272 SD
2019-2020
PSJA: 42 blocks, harmonic mean across blocks = 7.274103 --> MDES = 0.263 SD
|
After
We used the PowerUp! tool for power analyses (Shiny App here: https://powerupr.shinyapps.io/index/), allocating approximately 50% of students to the treatment and 50% to the control. We assumed that our model would include 20 covariates and would explain about 33.33% of the variance. Below are the minimum detectable effect sizes (SD units).
2018-2019 (same-year effects sample sizes)
District 1: 48 blocks, harmonic mean across blocks = 5.937367 --> MDES = 0.272 SD
2019-2020 (same-year effects sample sizes)
District 1: 42 blocks, harmonic mean across blocks = 7.274103 --> MDES = 0.263 SD
|