Study 1
Rogers and colleagues (2017) conducted a randomized controlled trial to evaluate the effects of communicating with guardians of school-aged youth about absenteeism rates in public schools in Philadelphia. Postcards contained either a simple message stating the importance of attendance or a more personalized message that stated how many days of school the guardian’s child had missed in the prior year. Postcards were mailed to eligible households with report cards in October of 2014. Eligible households included those that had a child who was enrolled in public school (grades 1–12) in the 2014–2015 school year and had any absences the previous school year (while they were in grades K-11 during 2013-2014). Households were not eligible if the guardians had opted out of receiving mail or had an invalid address, if the student had graduated, withdrawn, or otherwise was not enrolled as of June 2014, if atypically burdensome circumstances existed (students who were flagged as having a disability or who were homeless), if the student’s home language was different than the language of the consent form, if the student had participated in the pilot study during spring 2014, or if the student had perfect attendance in the year prior to the study.
The study targeted only one student per household. For households with multiple students, one student was randomly assigned to the student–guardian pair. There were 51,197 eligible students, about 40 percent of the total school population representing 217 Philadelphia schools and 1,290 unique school-grade combinations. This yielded an average of 40 students per school-grade combination. Of the 51,197 eligible students, 38,187 student-guardian pairs were randomly assigned to one of the following three groups: 1) The encourage condition group, which received one postcard that stressed the importance of attendance, guardian efficacy (guardian influence on attendance), and absence reduction as part of the guardian’s role; 2) the encourage + self-condition group, which received one postcard that had the same content as the encourage condition postcard and the number of days the student missed; and 3) the control group, which did not receive any mail. The total sample comprised 38,187 students. The encourage condition group (n = 12,756) was 50 percent female, 53 percent Black, and 69.4 percent were eligible for the federal school lunch program. They had missed an average of 13.17 days of school prior to the intervention. The encourage + self condition (n = 12,721) was 51 percent female, 53.7 percent Black, and 69.2 percent were eligible for the federal school lunch program. They had missed an average of 13.08 days of school prior to the intervention. The control group (n = 12,710) was 51.5 percent female, 52.8 percent Black, and 69.4 percent were eligible for the federal school lunch program. They had missed an average of 13.03 days of school prior to the intervention. There were no statistically significant demographic differences between the groups at baseline.
Changes in attendance were measured from attendance data provided by the school district. The attendance roll of participating students was limited to the period immediately after the postcards were sent (October 9, 2014) and before the district sent subsequent attendance mailings not included in the present study (December 31, 2014). Forty-three school days were included in the study period. The researchers fit an overall regression model and compared coefficients. The models included random effects, with students nested within schools, and fixed effects of grade level. Multilevel hierarchical modeling was used to determine whether the intervention had an impact on absenteeism. The researchers conducted subgroup analyses on absenteeism by grade level.
