Increasing school attendance in Seattle

What was the challenge?

The U.S. Department of Housing and Urban Development’s (HUD) strategic plan promotes educational advancement among HUD-assisted residents as a pathway to self-sufficiency. In the United States there are approximately 350,000 Public Housing units in which school aged children reside, presenting an opportunity to reach many students with education-focused interventions. Prior research has shown that school attendance is an important factor for both student achievement and high school dropout rates. In collaboration with HUD, the Seattle Housing Authority (SHA) and Seattle Public Schools (SPS) implemented three pilots during the 2018-2019 school year aimed at identifying effective strategies to increase attendance and reduce chronic absenteeism.

What was the program change?

SHA and SPS tested three distinct interventions during the 2018-2019 school year aimed at improving attendance and reducing chronic absenteeism. Each of the interventions focused on similar messages encouraging better attendance but differed in the mode of communication and the target audience. Although the original design included personalized communications with each student’s current number of absent days, complying with relevant privacy rules prevented data sharing that would allow for personalization, and the messages were more generic statements about the value being in class.

1. Robocall intervention: The first intervention, implemented by SPS, was sending a robocall the week before the beginning of the fall semester to households with students who had poor attendance the prior year. A senior leader within the school district recorded the message, which emphasized the importance of good attendance and the connection between attendance and instructional minutes.
2. Postcard intervention: The second intervention, implemented by SHA, was to mail households with SHA students either a letter or a postcard encouraging good attendance over the first 20 days of the school year. The goal of the intervention is to test which format was more effective at improving attendance.
3. Text message intervention: The third intervention, implemented by SHA, was to send a series of three text messages to households with SHA students during the spring semester. The text messages emphasized the importance of daily attendance and encouraged making plans to facilitate better attendance.

How did the evaluation work?

The three evidence-based interventions were tested with household-level randomized control trials. The sample for each trial are as follows:

1. Robocall intervention: Households with students who missed 5 percent of school days or more the prior year – including both SHA residents and other students – were randomly assigned to be sent a robocall (6,626 students in 4,458 households) or were selected into the control group that was not sent a robocall (6,487 students in 4,464 households). The robocall was dialed the week before the start of the fall 2018 semester. The primary analysis compares the total number of days absent and the percentage of days absent over the fall 2018 semester (between September 5, 2018 and January 29, 2019) for the two groups.
2. Postcard intervention: All SHA households were randomly assigned to be sent either a letter or a postcard at the beginning of the fall 2018 semester. The primary analysis compares the number of days absent over the first 20 days of the school year between the letter and postcard group.
3. Text message intervention: The third intervention, implemented by SHA, was to send parents of SHA students a series of text messages encouraging good attendance. Households with SHA students were randomly assigned to be sent the messages (2,424 students in 1,492 households) or not (2,364 students in 1,483 households). The primary analysis compares the number of days absent and the percentage of days absent during the spring 2019 semester (between January 31, 2019 and June 27, 2019) for the two groups.

What was the impact?

The results suggest that the communications did not change attendance outcomes.

1. Robocall intervention: Students in households sent the robocalls were absent on average 10.3 days during the fall semester, a reduction of 0.16 day from the control group. The difference is not statistically significant from zero. The average student in the robocall group was absent 13.1 percent of days he/she was enrolled, which was a decrease of 0.34 percentage point versus the control group, also not statistically significant.
2. Postcard intervention: We were not able to complete the analysis as planned. In cases where data are not available or the evaluation did not provide comparable comparison groups, we do not report results. In this case, outcome data were not available.
3. Text message intervention: SHA students in the text message group were absent on average 12.0 days during the spring semester. That is an increase of 0.18 day over the control group, but the difference is not statistically different from zero. Students in the text message group were absent on average 14.1 percent of days during the spring semester, an increase of 0.04 percentage point versus the control group. The difference is not statistically significant.

Verify the upload date of our Analysis Plan on GitHub.